The Development of Social Knowledge. Morality & Convention 0521273056, 9780521273053

Citation preview

Children are not simply molded by the environment; through constant inference and interpretation, they actively shape their own social world. This book is about that process. Elliot Turiel's work focuses on the de­ velopment of moral judgment in children and adolescents and, more

generally, on their evolving understanding of the conventions of social systems. His research suggests that social judgments are ordered, sys­ tematic, subtly discriminative, and related to behavior. His theory of the ways in which children generate social knowledge through their social experiences will be of interest to a wide range of researchers and stu­ dents in child development and education. Efliot Turiel is Professor of Education at the Universit:y of California, Berkeley

Cambridge Studies in Social and Emotional Development General editor: Martin L. Hoffman Advisory Board: Ni cholas Blurton Jones, Robert N. Emde, Willard

W. Hartup, Carroll E. Izard, Robert A. Hinde, Lois W. Hoffman,

Jerome Kagan, Franz J. Monks, Paul Mussen, Ross D. Parke, and Michael Rutter Over the last thirty years the study of child development has been in large part the study of cognitive growth. Recently, however, the focus has shifted. Not only developmental psychologists but social, clinical and educational psychologists, child psychiatrists, anthropologists. soci­ ologists, and students of animal behavior are actively engaged in explor­ ing the developmental roots of social behavior and emotional experi­ ence. The purpose of this series of books is to highlight the best new work and. further, to help bridge the present gaps in communication among disciplines and between basic and applied researchers.

�"O

CAMBRIDGE

t'i��

o O).,..:, M 1- w

UNIVERSITY PRESS www.cambridge.org ISBN 0-521-27305-6

Cover design by Richard Senior

111111 11 11

9 780521 273053

>

�

.·

UNIVERSIDAD PEDAGOGICA NfITRAL •

PROVfEOO�.

tlf!

o 1 q_�e

2 014·_ .

--......;¡

--

1/.Jn {t.I¡JYSt@éf

Structure and development

Interaction, construction, and structure Stages, developmental synchrony-asynchrony, equilibration Domains of knowledge and partial structures Methods of research

ISBN O 521 27305 6 paperback

INGRESO

vii

N a.c�11'!f___

3

Social experience and social knowledge

Morality and social convention Social interactions 4

Dimensions of social judgments

Criterion judgments, justification categories, and social events Research on criterion judgments Criterion judgments, justification categories, and familiarity with the issue Discontinuity or continuity in the domain distinction

:i.Q}:}:¡_b"'"t""O..__�-----i 5

Rules and prohibitions

Research on concepts of social rules The systematic nature of social rule concepts 6

1 2 4 8 9

15 20 21 33 34 40 50 52 55 63 68 75 81 97

The development of concepts of social convention and coordination of domains

lOO

vi

Contents Affirmations and negations: equilibrium and disequilibrium in concepts of social convention Af f irmation-negation and developmental transformations Domain combinations

7

The deve1opment of moral judgments

Classification of events and issues used as stímulí Analyses of the development of moral judgments Elements of the heteronomous orientation Does heteronomy precede autonomy? Modífications in analyses of heteronomy Age-related changes in moral judgments 8

Superego formatíon Behaviorístic-internalizatíon approaches Social judgments and actions: coordination of domaíns

Research on the relations of moral judgment and action Consistencies and inconsistencies in judgment and action Domains ofjudgment and action 10

1 12 114

Conclusions: Interaction, development, and ratíonality

Ratíonality and nonrationality in social domains Categories of social knowledge References Index

Preface

130 131 136 137 145 153 157

Noncognitive approaches to moral deve1opment: intemalízatíon and bio1ogica1 determinísm

9

102

16 1 166 173 187 188

The deve1opment of social knowledge in children and adolescents is a broad topic requiring sorne choices regarding the focus of analysis. Those choices are

naturally guided by one 's theoretical orientation to social reasoning and its for­ mation and transformatíons in the process of growth. My orientation, which is structural and developmental, has led me to a focus on categories of social

reasoning. Much of the research for the theory presented in this book had its origins, about ten years ago, in close analyses of delimited domains of social knowledge. The research was aimed at disceming the differences and similarities in moral judgments and concepts of the conventions of social systems. lt quickly

193 201

became apparent that moral reasoning and concepts of social organization consti­

211

social world.

tuted two fundamental and separable aspects of children's understanding of the

2 14

In the intervening time my research on morality and convention has gone

222

beyond the original focus to include other importan! aspects of social develop­

227 237

the framework for an understanding of children's social interactions with peers,

ment. Moreover, the analyses of categories of social knowledge have provided their relations with adults, their processes of development, and the relationship between social judgments and actions. A central thesis of this book is that social life is guided by rational processes and that the construction of fundamental and distinct categories of social knowledge begins in early childhood. The idea of distinct categories of reasoning does not imply a dualism of thought and action or of individual reasoning and cultural content. Rather, it is proposed that the separation of domains of knowledge provides a basis for understanding the interrelations among thought, actíon, and cultural content. The theory of social development is interactive. lt propases that social development evolves through children's reciproca] interactions with the social environment. In part, this book is aimed at brínging together the various components of the research programs undertaken during the past ten years. Sorne of the research findings are being presented for the first time. There is also extensive discussion of relevan! findings from a set of previously published studies. Much of that vii

Prejace

viii research was conducted by colleagues

(in

collaborative and in independent ef­

forts) who have shared an interest in issues related to the development of do­ mains of social knowledge. They include Allen Black, Philip Davidson, Larry

Nucci, and Judith Smetana.

The book has two main goals. One is to further understanding of how people

1

lntroduction: Approaches to the study of social knowledge

reason about morality and convention and how those domains of reasoning develop from early childhood through adolescence and early adulthood. The second goal is to provide - through extensive and detailed analyses of morality and convention - documentation for the proposition that individuals form stable systems of social knowledge that are organized around domains. In this sense, the present volume is meant to build an empírica! foundation for sorne other theoretical propositions to be presented in a subsequent vol ume. To achieve these goals, the book is organized in the following way. I have begun with defini­ tions, hypotheses, and theoretical assumptions regarding social knowledge and development. This is followed by discussion of research findings bearing on the hypotheses and propositions. Last, there is a consideration of the implications of this research for a general view of social development. I would like to acknowledge my gratitude to those who have assisted me in the

?f

This book is about the ways individuals develop social knowledge through their interactions with the social environment. Knowledge about the social world is, in large measure, socially derived knowledge. The proposition that social knowl­ edge is socially derived, however, can be interpreted in two very different ways. 1t can be ta.ken to mean that it is knowledge transmitted to the individual by other persons, so that the knowledge acquired is dependent on what is transmitted, or it can be ta.ken to mean that it is knowledge constructed by individuals specifically

this book. The early stages of the research were facilitated by a

about social phenomena. The analyses presented in this book are based on the

fellowsh1p from the John Simon Guggenheim Memorial Foundation. The actual

latter interpretation of the social derivation of social knowledge. The analyses

writing

writing of the book was aided by a fellowship period al the Van Leer Jerusalem

document that, starting in early childhood, individuals form conceptual systems

Foundation in Israel. The Foundation provided time and a very stimulating . settmg for thought and writing. Thanks are due to Raphaella Bilski and Yehuda

for interpreting the varied elements of their social experiences. Social develop­

Elkana at Van Leer. Throughout the past few years I have been given research

world, by making inferences and forming theories about experienced social

support by the Institute of Human Development at the University of California,

events.

Berkeley; I thank the past director of the Institute, Paul Mussen, and the present

ment is a process by which individuals generate understandings of the social

The study of social development requires two interrelated analyses: the nature of realms of social interaction and the explanation of processes of acquisition or

director, Ed Swanson. Paul Mussen also provided helpful comments on an earlier version of the

development. Social scientists have extensively considered development and

manuscript. I also thank Carolyn Hildebrandt, Melanie Killen, and Larry Nucci

categories of social interaction, culture, and society. However, each of the

for their comments. Finally, thanks to Helen Clifton for her excellent work in

concerns has been dealt with by separate social scientific disciplines. The most

typing and retyping the manuscript.

extensive and explicit investig�tions of individual social development, as would

Berkeley, California

1897; Freud, 1923/1960, 193011961; Miller & Dollard, 1941; Piaget, 1932, 1951!1962; Skinner, 1971; Watson, 1924) have attended to

be expected, come from the discipline of psychology. The major psychological October

1982

Elliot Turiel

approaches (e.g., Baldwin,

the explanation of the means by which children acquire social behaviors and social knowledge. Among social scientists, it is mainly anthropologists and sociologists who have studied systems of social interaction through analyses of culture and society. For their part, students of culture and society ha ve generally disregarded systematic study and the forrnulation of explanatory principies of psychological functioning and development. Instead, the tacit - and sometimes not so tacit- assumption is made that individuals are molded by and come t o fit

2

The development of social knowledge

their social contexts. The assumption is that individuals accommodate to the patterns or elements of the culture or social system, thereby coming to mirror the culture in their personalities, thinking, and behavior (Benedict, 1934, 1946; Geertz, 1973; Shweder,

1981; Whiting & Child, 1953).

A concern with acquisition, however, must go beyond assumptions to direct investigations and explication of the process of development. The nature of the individual's relation to the broader social system, the ways in which behavior is influenced by social experiences, and the genesis of the individual's social makeup are al! issues requiring explanation. Are individuals shaped by the social world, or do they develop conceptual systems for understanding and transforrning the social world? Analyses of culture and social systems require corresponding anal­ yses of psychological-developmental processes if one is concerned with the integration of individuals into such systems. In turn, psychological investigations of development need to be inforrned by analyses of categories of social interac­ tion. The study of the individual's social development, thinking, and behavior requires identification of the types of social elements typically made by students of culture and society, including categories like economic, political, religious, moral, and social organizational systems. The research described in this volurne documents that individuals develop theories about categories of social interaction. Those theories, however, are not simply about each specific or isolated social interactional element experienced. The elements are organized into more general categories of thinking. But social interactions are not conceptual!y organized to encompass all of social thinking into one system. A main proposition of this volume is that social judgments are organized within domains of knowledge. A related proposition is, therefore, that the study of social development requires classification and identification of do­ mains of social judgment. One of the tasks, then, is to study the organizations and reorganizations of thinking within domains of social judgment. The meaning and boundaries of domains of thinking are considered throughout this volume, and especially in Chapter

2, where the general theoretical framework for the

study of development is considered. The theoretical perspective is a structural one that is most closely associated with the work of Piaget. Many other re­ searchers have also taken a structural approach to development, including a recently burgeoning interest in the development

m·

social judgments.

3

Jntroduction

of this volume is on two social domains: convention and morality. Social conven­ iions are behavioral uniforrnities that serve to coordinate social interactions and are tied to the contexts of specific social systems. Conventions are based on arbitrary actions that are relative to social contexts. Through their participation in social groups, such as the family, school, or with their peers, children form conceptions about social systems and the conventions, the shared expectations, that coordinate interactions. Whereas conventions are deterrnined by the social system in which they exist and constitute part of the definition of the social system, the moral domain refers to prescriptive judgments of justice, rights, and welfare pertaini g to how peo le � � ought to relate to each other. Moral prescriptions are not relattve to the soc1al context, nor are they defined by it. Correspondingly, children's moral judgments are not derived directly from social institutional systems but from features inher­ ent to social relationships - including experiences involving harrn to persons, violations of rights, and conflicts of competing claims. This volume presents detailed analyses of morality and social convention and a theoretical point of view on the development of social judgments and, in particu­ lar, on a domain-specific interpretation of social reasoning. After considering the general principies of structure and development (Chapter 2), definitional criteria for the two domains are presented in Chapter

3. As discussed in Chapter 2, a

structural approach propases that thought and action are closely linked with each other. The links are twofold. The development of thought stems from the chíld's actions upon, and ínteractions with, the environment. Whereas action is the source of conceptual development, socialjudgments- once they are forrned- are not distinct from actions manifested in behavioral situations. Relations between thought and action are considered first i n Chapter 3 and la ter in Chapter

9.

Chapter 3 includes analyses of the experiential sources of social judgments and describes a series of observational studies of children 's social interactions in naturalistic settings. Before dealing in Chapter 9 with the relation of social judgments to behavior, severa! chapters present analyses of the organizations of moral and conventional thinking. The conceptual features of each domain and the characteristics that serve to distinguish them are considered in Chapte¡s 4 and 5. The data presented in those two chapters offer evidence that morality and convention constitute distinct and non-age-related domains. Developmental changes are apparent within

Domains of social judgment The domain of social interaction has a· significant bearing on the individual's thinking, action, and development. The theoretical and empírica! analyses pre­ sented in this volume address the question of the relations between social interac­ tion categories and the development of social judgments and actions. The focus

each domain; Chapter 6 deals with the development of concepts of social conven­ tion and Chapter

7 with the development of moral judgments.

The analyses and documentation presented in this volume provide the founda­ tion for extending, in a subsequent volume, the domain-specific interpretation of structure and development to a more general theoretical framework regarding social development. The broader framework, which encompasses morality and

4

The development of social knowledge

convention, is based on the identification (Turiel,

1979, in press) of three general

and fundamental categories of social-conceptual knowledge, as well as the dif­

Jntroduction

5

The concept of culture is still less than clear or precise, still evolving, and, as with most central concepts, unanimity does not exist within the discipline regard­

ferentiation of social-conceptual knowledge from social-informational knowledge.

ing its meaning or functions. Differences of interpretation exist at general levels

On the basis of the assumption that the individual 's social world includes other

of analysis. Sorne anthropologists view cultures as homogeneous, integrated

persons, relations between persons, and institutionalized systems of social inter­

wholes (e.g., Benedict,

action, it is proposed that the child's structuring of the social world revolves

Geertz,

around three general categories. These are ( l ) concepts of persons or psychologi­

small communities, are characterúed by diversity and elements that are not

cal systems (the psychological domain),

(2) concepts of systems of social rela­

tions and organizations (the societal domain) - of which convention is but one component, and

(3) prescriptive judgments of justice, rights, and welfare (the

moral domain).

1934, 1946; Whiting & Child, 1953). For others (e.g., 1973; Schwartz, 1978; Shweder, 1979a,b, 1980), aiJ cultures, including

necessarily integrated with each other: "Even the most minimal community would reveal a ... distinctiveness of personal and cultural variation which, when viewed against the scale of the culture and by the concems of its members cannot be considered trivial" (Schwartz,

1978, p. 420). Within either the integrated or

the diverse view, culture has been regarded as performing controlling functions. Social understandings or social control This book is also about the individual' S relation to and interaction with the social environment, how to characterize those interactions and, especially, how those interactions influence development. The starting point for the analyses of moral­ ity and social convention is the premise that through their social experiences children develop ways of thinking, or theories, about the social world. One type of theory central to social development pertains to moral issues. Children also form theories of social organization, through which they understand the meaning and function of conventions. In coming to understand social systems, people act Iike social scientists, attempting to observe regularities and explain their exis­ tence. Correspondingly, individuals attempt to understand and critically or posi­ tively evaluate the sources, functions, and justifications for the rules they are expected to obey. In other words, social life is not detached from thought 1•

processes. Indeed, in a long tradition of research, judgment or cognition has been regarded as central to social development and behavior (Asch,

1952; Heider, 1958; Kohlberg, 1969; Kohler, 1938; Lewin, 1935; Mead, 1934; Piaget, 1932; Vygotsky, 1934/ 1962; Werner, 1957; Wertheimer, 1935). A tradition of research on the development of moral judgments was originally pioneered by Piaget

(1932) and extended by Kohlberg (1963, 1969) sorne years later. However, the cognitive perspective of social domains is not shared by al! students of development or of social systems. A contrasting view of morality and convention is that they rcpresent variants of externally determined standards, which serve to shape and guide people's behaviors, as well as delimit or control their choices. From the societal viewpoint, morality and convention, which are not always distinguishable, serve a controlling function, forcing individuals to accommodate to the group. As mentioned eariier, sorne anthropological perspectives posit the primacy of culture and assert that culture serves to exert control over individual behavior.

As an example, the thesis that cultures are homogeneous, integrated wholes was posited by Benedict

(1934) in her analysis of Patterns of Culture. For

Benedict, the coherence of culture produces consistency across domains. The type of integration found in one culture, however, is likely to differ from that of another. Moreover, because a cultural pattem is posited to be intemally consis­ ten!, the practices of one culture cannot be compared with those of another. Benedict's type of relativism stems from the idea that the practices of each culture mak:e sense with regard to its form of integration. Benedict maintained that all social regulations are culturally determined customs, whose meaning and importance could be understood only in relation to their role in a cultural pattem. The unity of culture is replicated in the individual. It guides the ways children are brought up, so that they come to manifest in their personalities and behavior the pattems of culture or what is sometimes referred to as national character. In

Chrysanthemum and the Sword, for instance,

The

Benedict describes child training in

Japan as successfully geared to the formation of a Japanese adult whose personal­ ity mirrors the culture. This was consistent with her earlier premise that the individual's social behavior reflects a nonrational incorporation of the standards of the culture: "The life history of the individual is first and foremost an accom­ modation to the patterns of standards traditionally handed down in his commu­ nity" (1934, pp.

2-3).

The proposition that individuals are controlled by the social environment is not always associated with a holistic or integrated view of culture. The idea that the individual is shaped by the social environment has been combined with the view that culture is a series of mechanisms of b ehavioral control. As Geertz (1973, p.

44)

put it, "Culture is ...a set of control mechanisms - plans, recipes, rules,

instructions (what computer engineers cali 'programs') - for the governing of behavior ... Man is precise!y the animal most desperately dependen! upon such extragenetic, outside-the-skin-control mechanisms, such cultural programs, for ordering his behavior." Conceived in less holistic terms than by Benedict, cul-

6

The development of social knowledge

ture is, nevertheless, the means by which the behavior of individuals is con­ trolled. The individual requires a narrowing of behavioral possibilities, a channeling of capacities into a social order: "Undirected by cultural pattems- organized systems of significant symbols- man's behavior would be virtually ungovem­ able, a mere chaos of pointless acts and exploding emotions, his experience virtually shapeless" (Geertz, 1 973, p. 46). A goveming thesis of these approaches is that culture is primary and dominates the individual. lt is assumed that a correspondence exists between individual behavior and what exists in the culture. The presumed correspondence between efforts at social control and individual behavior should not go unquestioned; the extent of correspondence is neither self-evident nor empirically verified. Whether one assumes a greater or lesser extent of correspondence, however, it does not fol!ow that social control is the causal mechanism for the formation of social behavior. In addressing the ques­ tion of social psychological development, a series of issues must be posed and answered with specificity and detail. First is consideration of the nature of the person's inherent capacities and orient�tions. Are they such that living in a social world results in narrowing and channeling, so that "the breadth and indetermi­ nateness of his inherent capacities are reduced to the narrowness and specificity of his actual accomplishments" (Geertz, 1973, p. 45)? Or are they such that social life results in a broadening of social interactions and expansion of knowl­ edge? Although the concept of control is applicable in the former case, it is . inappropriate in the latter. The nature of. the individual's inherent capacities is not unrelated to a second issue: the nature of the individual's interaction with the social environment. Is the emphasis to be placed upon the control exerted by society, or upon the individual 's attempts to understand and manipulate his or her environment, or upon the point of intersection between the two processes? In tum, we must inquire into the nature and courses of change that individuals undergo: ls social development best represented, as Benedict put it, as an ac­ commodation to the cornrnunity, or is it the construction of modes of reasoning about social categories? Does it involve an increasing accumulation of enviran­ mental content or a progression of qualitative changes in mental-psychological organization? These questions are central to developmental analyses. Answers to sorne of these questions are implicit in theorizing about social and cultural systems that assumes causally imposed environmental control over the individual's behavior. Such implicit theories are, in a general sense, consistent with sorne explicit theories of social development- most notably, those of behavioristic and psy­ choanalytic theorists (comparisons of these noncognitive orientations with the cognitive ones are made in Chapter 8). lt may be said that for behaviorists social development is under environmental control, whereas for Freudians it is under cultural (or societal) control. Behavioristic theories have explained the develop-

Introduction

7

ment of morality through the psychological mechanisms of conditioning and the 1eaming of values and rules. Morality i s thus seen as a combination of fixed habits and direct mental representations of values and rules. Freudian theory has explained moral development as the formation of a conscience (superego), an intemalízed mental agency whose function is to control, regulate, and transforrn ínstinctual impulses that conflict with societal functíons. Although there are important differences between behavioristic and Freudian theories, they share the assumption, along with the theories of culture discussed earlier, that the social environment serves to regulate and control human con­ duct. From this assumption three tenets follow: ( 1 ) a dichotomy exists between the individual and the group; (2) a dichotomy exists between self-interest and other-interest; and (3) nonrational or irrational processes predominate in deter­ mining behavior in the social domains. In these conceptions the individual is regarded as entering the world with impulses, needs, drives, or, according to Freud, with complex pattems of instincts. Whichever way it is put, in analyses of social development this translates into a view of a child whose initial state of selfishness and impulsiveness must be curbed and made social. The developmen­ tal process is one in which originally externa! content residing in the group (in the form of standards, regulations, prohibitions, etc.), becomes incorporated by the child. That is, with increasing age the child's behavior is modified or socialized so as to render it more consistent with group interests and standards. Socializa­ tion is aimed at modifying the child's natural self-interested tendencies into group attachment and altruistic tendencies. The proposed dualities between the individual and the group and self-interest and altruism are rooted in the view that the social domains are primarily nonra­ tional. Social behavior is, in the main, guided by emotions; reason is, at best, secondary. It is needs, drives, and passions that motívate the child's initial interactions with the social world. It is rewards, punishments, and threats, along with their resultant fears, anxieties, and ambivalences, that cause the shift from self-centeredness to greater sociability. Indeed, if the social environment only serves to regulate and control, it does follow that the social domains are nonrational. However, the social environment does not merely exert control upon individuals but is understood by them. Paral­ leling the diversity of culture is the diversity of the individual'S understanding of the social world. Development entails an expanding social knowledge and an increasing complexity of social interactions. The child is not motivated solely by needs and drives that conflict with social obligations and that make for a dichot­ omized relation of the individual to the group. Rather, the child is in a reciproca! interaction with his or her environment, through which descriptive and prescriptive concepts are constructed.

9

Structure and development

2

Structure and development

tion (1923), and moral judgment (1932). Undoubtedly, the early research has had a substantial influence upon recent concems with social cognitive develop­ ment. However, more important influences - particularly for the point of view and the research fmdings presented in this volume- have stemmed from Piaget's Iater theoretical forrnulations regarding thought, structure, and development. A distinct shift was evident in Piaget's approach starting with his work on sensori­ motor development (Piaget, 1936/1963, 1952). In addition to an exclusive con­ centration on nonsocial cognition in his later work (e.g., logic, mathematics, physical relations, causality, probability), Pi�get formali�ed his struc ral and developmental conceptions. Whereas the earher formulattons charactenzed the child's thought as progressing through general and global phases, his later formu­ lations included a series of distinctions serving to differentiate types of knowl­ edge. For instance, distinctions are drawn between logical-mathematical concepts and physical concepts, as well as between operative and figurative fonns of knowledge (Turiel, in press). Moreover, severa] researchers (e.g., lnhelder & Sinclair, 1969; Inhelder, Sinclair, & Bovet, 1974; Kohlberg, 1969; Langer, 1969, 1974, 1980; Strauss, 1972, 1981; Sugarman, 1979), along with Piaget, have provided systematic formulations of structural development. Those principies of development are outlined in this chapter, without an exten­ sive or comprehensive explication. The purpose of this discussion is to provide an overview of the basic concepts and to serve as a general guide to the presenta- . tion of research procedures, findings, and interpretations in the subsequent chap­ ters. Of course, the remainder of the volume ihcludes further extensions of the theoretical framework.

�

The analyses presented in this volume are based on the assumption that individ­ uals define, interpret, and judge social relations. The individual's interactions with the environment are characterized by efforts to understand other persons and relations among persons. Hence, the individual is in a reciproca! interaction with the social environment. The research discussed here was designed to examine the social judgments of children and adolescents, the developmental course of those judgments, the relations of social experience and development, and the relations of social judgment and social behavior. The developmental approach to social judgment currently represents one of two main trends in psyc!tological research emphasizing the role of cognition in the social domains. Within the subdiscipline of social psychology, attribution theorists (e.g., Jones & Davis, 1965; Kelley, 1967, 1973; Nisbett & Ross, 1980; Ross, 1977) have been primarily concemed with causal inferences in the individ­ ual 's predictions and conceptualizations of the behavior of other persons. Devel­ opmental psychologists have paid relatively little attention to childreri's understandings of the psychological attributes of persons, concentrating to a greater extent on the study of children' s concepts of social relations (potential relations between the two approaches are discussed by Ross, 1981, and the place of the psychological domain in developmental forrnulations is discussed by Turiel, in press). The developmental approach to.social cognition has its origins in a related, but somewhat different, tradition of structural psychology from the origins of attribution theorists. Whereas attribution theorists were influenced by Gestaltists (e.g., Asch, 1952; Lewin, 1935) concemed with social psychological problems, the recent work in developmental social cognition draws heavily on theories concemed with the relations between structure and development, such as those of Baldwin (1906, 1915), Vygotsky (193411962), Wemer (1937, 1957), and especially Piaget (1929, 1932, 1936!1963, 194711950, 1970a,b). Piaget conducted some extensi ve research on children's social reasoning dur­ ing what may be considered an early, exploratory phase of his work. lncluded among those early studies were investigations of perspective-taking, communica8

Interaction, construction, and structure

An overriding premise of the structural approach is that thought is organized and that it is constructed out of the child's interactions with the environment. The meanings of construction and interaction and their interrelations in a develop­ mental context have been summarized by Piaget ( l970a, p. 704): In

order to know objects, the subject must act upon them, and therefore transfonn them: he must displace, connect, combine, take apart and reassemble them. Fro � the most elementary sensory-motor actions to the most sophisticated inteiiectual operat ?ns, wh1� h

t

are

interiorized actions carried out mentally, ...knowledge is constantly hnked w1th actions or operations, hat is, with transformations....Knowledge, then, at its origin

;

neither arises from objects nor from the subject, but from interactions ...between the subject and those objects.

The initial sources of the construction of thought and knowledge are the child's sensorimotor and interiorized actions upon, and interactions with, ob­ jects, events, and persons. The thesis is that thought and knowledge are neither

10

The development of social knowledge

11

The psychometric view of intelligence will be considered in the context of a

givens in the biological makeup of the individual (the subject), nor do they stem directly from the environment (objects). The thesis, therefore, is interactional and not maturational. However, over the years the proposition that thought is

discussion of methods of study. For now, the point to be made is that from a

constructed through individual-environment interactions has often been misin­ terpreted to be a nativistic or maturational one. As early as 1930 Isaacs (p. 58)

process of cognitive activity. According1y, the concept of intelligence is not characterized as a quantitative attribute of persons, but as systems (or forrns of

critically stated her view that Piaget's

organization) of thought, which in ontogenesis undergo transformations, that is, constructions. It is indeed true, as Wilson wrote, that Piaget referred to the

views on the development of the child's language and thought , judgrnent and reasoning,

and conception of the world, constitute the most arresting modern staternent of the theory

o f rnaturation . . . . But Piaget's conclusions are, I think , lessened in their final value because he does no! use the concept of maturation (which he expresses as the structure of

the chi ld' s mind at different ages) sparingly enough nor critically enough.

More recently, Wilson ( 1978, pp. 68-69) praisingly observed that Piaget spent a lifetime charting the often s urpris in g stages chi l dren pass through in their more

purel y intellectual growth. . . . lt is no coincidence that he calls this conception 'genetic episternology', in effect the study of the hereditary unfolding of understanding .

In all likelihood, this assimilation of the hypothesis of interaction and con­ struction to a maturational one stems from the often-made assumption that chil­ dren's development, and behavior in general, are causally determined by two

factors, heredity and environment. If development is not explained as a direct function of experience, it is assumed that one must be positing a nativistic or maturational explanation. Among those who take innateness and experience as the sole factors, the disagreements are over the relative emphasis placed on each factor. Most ha ve regarded innateness and experience as representing dual causes of behavior. However, so me place the greatest emphasis on in nate factors (e. g . , instincts, predetermined structures of behavior or ideas, maturationally deter­

,, ,

Structure and development

mined capacities), others regard the acquisition of environmental content as primary, and still others attribute more or less egua! weight to the two factors. A paradigmatic example is provided by the long-standing debate over the

structural perspective intelligence is viewed not as a capacity or a trait but as a

structural theory of intellectual growth as genetic epistemology. However, the use of the term "genetic" is meant to connote development and not a hereditary unfolding. In its developmental sense the term has been used in two contexts, genetic epistemology and genetic psychology. Genetic epistemology is not re­ stricted to individual development, but refers to the wider sense of the growth of all knowledge including, for example, branches of scientific disciplines. In any case, the developmental connotation of the term "genetic" has been explicitly stressed by Piaget and Inhelder ( 1969, p. viii): To dispel any arnbiguity about terrninology, Jet us note first that the word 'genetic', as

used in the expression 'genetic psychology' , was introduced by psychologists in the

second half of the nineteenth century to refer to the developmental aspects of psychology . Later, biologists began to use the term 'genetics' in a more restricted sense. In the current language of bi ologists, 'genetics ' refers only to the rnechanisms of here dity and does not n i clude the study of ernbryogenetic or developmental processes. The term 'genetic psy­ cholog y ' , however, continues to refer to individual developmen t (ontogenesis) .

This is not to say that there is no biological basis to deve1opment. It is to say that organizations of knowledge are not proposed to be preformed in the nativis­ tic sense. The biological or inherited factors contributing to the development of thought are, in addition to general capacities for cognitive activities, the dual functions of assimilation and accommodation (Piaget, 1 936/ 1963, 1 970a, 1980a) .

determinants of human intelligence waged by those using psychometric mea­

Assimilation and accomrnodation are inherent to interactions between biological systems and the environment; through assimilation the individual modifies events

surement (the familiar IQ tests). Within that tradition, intelligence is defined as a trait or capacity possessed by individuals to a greater or lesser degree. Psycho­

to fit his or her structures and through accommodation the individual's structures are modified to fit events. Moreover, assimi1ation and accommodation are func­

metric measures of intelligence are presumed to pro vide a reliable es timate of the

tions that operate in individual-environment interactions at every level of devel­ opment, including the activities of infants and adults. The process of assimilating

amount of intelligence possessed by an individual (relative to others). Competing explanations of the sources of intelligence vary usually with regard to the degree to which it is determined by innate capacities or by experience. Sorne have

environmental events can be seen, for example, in the infant's attempts to suck objects like rattles or blocks, thus fitting the objects to a sucking schema. In

maintained that intelligence is primarily a genetically determined capacity only

addition to assimilation, experienced events produce accommodation of existing

minimally affected by experience. Others have argued that, in spite of sorne

schema to the structure of the object. Infants accommodate their sucking schema

contribution of genetic traits, intelligence is primarily determined by experience.

to the type of object; they suck the nipple and thumb in different ways, and b1ocks become nonsuckablc objects. The concepts of assimilation and accommodation imply a reciproca! relation

Experience is said to account for most of the individual differences in the amount and content of acquired knowled e . (Sorne have also urged for a more balanced

g

contribution of the two factors. )

between the subject's structuring activities and experienced events. That is, it is

1

·

12

The development oj social knowledge

necessary to analyze responses to stimuli in their reciproca! fonn. Responses are not detennined entirely by externa! events, such as the characteristics of stimuli or associated positive and negative reinforcements. As an illustrative example of how a response entails a structuring of the stimuli eliciting it, consider the classic experiment conducted by Watson and Rayner (1920). The experiment, which presumably demonstrated the conditionability of emotional reactions in infants, was based on the hypothesis of a unilateral connection of stimulus and response (S

---"t

R ) . Watson and Rayner claim to have conditioned a fear of rats in the

infant by pairing the appearance of the rat with the onset of a loud noise (the durability and generalization of the response to other white, furry animals is open to sorne question; see Harris, 1979). Watson and Rayner maintained that pairing an aversive stimulus with an originally neutral stimulus was sufficient to produce the fear response. Presuma­ bly, the source of the fear response (the rat) was initially arbitrary and was made nonarbitrary by the conditioning process. However, the results of experiments (Bregman, 1934) using the same conditioning techniques with different stimuli have shown that the relation of the subject to the object contributes to the leaming that occurs. Using rnethods similar to those of Watson and Rayner, Bregrnan attempted to condition infants to fear wooden objects and cloth curtains

Structure

and development

13

The research focused o n part-whole transformations: the transformation of ob­ jects from one form to another, and the formation and deformation of objects into parts and sets. The infants were presented with a small series of different objects (e .g., rings, columns, and spoons) made out of nonrnalleable and malleable

materials. The analyses were based on detailed recording of the spontaneous activities of infants , as well as their reactions to specific manipulations o f the objects by the experimenter. The logical operations found to be part of that early period of cognitive development consist of combinativity (e.g . , cornposing and decomposing), relational operations (e. g . , addition and subtraction), and condi­

tional operations (e.g., negation and correlation). The cognitions of this period represent only the beginnings of later constructions. Indeed, they have been termed (Langer, 1980, l 9 8 la) "proto-operations" and characterized as "proto­ logic" to signify that they are precursors to the later logical operations and inference structures of childhood (lnhelder & Piaget, 1964) and adolescence (Inhelder & Piaget, 1958). This is not the place for an extensive discussion of the sensorimotor origins of physical cognitions and logical operations. The brief mention of findings from research on infant cognition serves to illustrate that even at the youngest ages, the individual' s interactions with the environment stem frorn organized systems of

of various shapes and colors. Bregman's experimental efforts were unsuccessful: She was unable to condition fear responses to the inanimate objects.

thought: "The structures of physical cogniti'on are of a piece; they constitute a

The Bregman and Watson and Rayner findings demonstrate that even at the

mathematical cognitions take one fundamental form- operations" (Langer, 198 1 b,

leve! of experimental efforts at conditioning fear responses in infants a reciproca] relation exists between responses and stirnuli. The findings can be interpreted to mean that the nature of the stimulus, in concert with the child's schema, are significan! determinants of responding and learning (Nisbett & Ross, 1980). From the point of view of an infant, anirnate objects are more likely to have fear-evoking qualities than a block of wood or piece of cloth. Development during infancy, however, goes well beyond the leaming of discrirninative emo­ tional responses . During infancy, or the sensorimotor period of development (Piaget, 1936/1963, 1952, 1954), the initial cognitive organizations are con­ structed that lead to the fonnation of representational thought and symbolization. Piaget has traced the development of infants' means-ends activities, through which their actions are coordinated with desired objectives . The focus of that research was on the early constructions of physical knowledge, that is, knowl­ edge of objects (e. g . , the onset of object pennanence), spatial relations, temporal order, and causality. Recently, sorne importan! discoveries were made by Langer ( 1 980) regarding the origins of logic. In extensive, detailed, and precise research on infants' (6 to 1 8 months of age) manipulations of different types and shapes of objects, evi­ dence has been obtained of the early formation of elementary logical structures.

unified organization of one fundamental form - functions. Sirnilarly, logico­ p. 16). It is proposed, therefore, that thought is characterized by systems of organization in which the elements or parts are subordinated to the laws of the whole (Piaget, 1 970c; Turiel, l978a). However, it should be clearly noted that al! aspects of thought do not constitute one general systern of organization. Even during infancy distinctions are drawn between physical cognition (means-ends activities that are goal-directed; functions) and the precursors to logical-mathernatical cognitions, which entail rudimentary fvrms of classification and seriation in part-whole transformations. Moreover, distinctions are drawn between aspects of knowledge that form organized systems and those that do not. (More will be said shortly in this chapter about the boundaries of systems of organization, which is a major topic of this volume with regard to social cognition.) It is also proposed that development entails systematic sequential changes in the organiza­ tion of thinking within domains. As indicated, the cognitions of the sensorimotor period are the antecedent structures of later constructions of physical and logical cognitions. A developmental sequence represents qualitative changes in ontogenesis fro m simpler to more complex fonns of organization. The child's interactions with the environment are based on the existing organizations of thought (stages or levels of development). That is, relations between the individual and the environment

14

The development of social knowledge

are interactional in the sense that events are interpreted through the individual's

the disconfirming studies were criticized and the confirming studies were viewed

structures of thought. One example of how this is manifested comes from the

as superior.

multitude of findings showing that children at different developmental levels

From the attribution perspective of the researchers who conducted this experi­

respond to the same intellectual tasks in very different ways. It has also been

ment, the assimilation of new information to existing theories is termed "belief

found that leve! of development even sets limits on the child's comprehension of

perseverance" (Nisbett

input from the physical and social (e .g. , in the form of communications or

stating that individuals have coherent systems of thought that serve to structure

instructions) environments. Examples come from experiments on children ' s moral

their interpretation of new information and evidence. Accordingly, similar types

judgments (Rest, 1973; Rest, Turiel,

,..

15

Structure and development

&

Kohlberg ,

1969). In these experiments

&

Ross,

1980). Belief perseverance is another way of

of structuring of information or environmental input has been documented in

measures were obtained of children ' s comprehension of presented solutions to

social psychological experiments with adults and children. Other related phe­

problems that corresponded to levels different from their own. The degree of

nomena central to a structural approach to development have been documented

comprehension shown by a child was related to the match between his or her own

by attribution analyses (Ross,

&

1977), as well as by the work of cognitive psy­ 1972, 1973; Tversky, 1977; Tversky &

Tversky,

leve) of thinking and the leve) of the communication. Children were able to

chologists (Kahneman

comprehend communications at levels below their own and at their own leve!.

Kahneman,

For the most part, they were unable to comprehend communications at Jevels

children's thinking has been the analysis of their conceptual errors . The best­

above their own. In fact, in many cases higher-level communications were

known example, but it is only one example, is Piaget's analyses of young

reinterpreted to resemble the child's own leve! of thinking or, especially when

children's judgments of nonconservation of number, length, and weight. On a

the communication was discrepant with the child ' s position, to resemble levels

variety of tasks (classification, seriation, transitive relations, chance), children ' s

less advanced than the chi1d's own.

errors i n their solutions o f tasks provide a data base for descriptions o f levels o f

1974). Starting with Piaget, an important means for understanding

Perhaps the most striking examp1es of how the individual' S existing modes of

cognitive operations. The formulation o f a developmental sequence includes

thinking have a powerful influence on the interpretation of new information

both analyses of earlier levels of thought and their relation to the more adequate,

come from research having an attribution-theoretic perspective. One example

more powerful conceptualizations of later levels of thought. Similar!y, with

(1979 ), who examined

regard to adult judgments, research by Ross (1977) has revealed errors in psycho­

how adults deal with empirical evidence that is confirming and disconfirming of

logical attribution , and research by Kahneman and Tversky has focused on

comes from an experiment by Lord, Ross, and Lepper

strongly held beliefs . ConflfiJling and disconfirming evidence was presented to

systematic errors in scientific and statistical inferences. Although the tasks used

subjects who had expressed either the belief that capital punishment was a

in the research with adults are of greater difficulty and complexity, the pattems

deterrent to potential murderers or that it did not serve as a deterrent. A compli­

of analysis are similar to the developmental research in that judgments are

cated experimental design was used that may, for present purposes, be summa­

analyzed for underlying strategies, and (2) compared with normative standards

rized as follows. Participants in the experiment were told about the results and

(in this case, of scientific inference).

(1)

methods of two studies (made· up by the experimenters) on the deterrent effects of capital punishment. The two methods provided were ( 1 ) comparisons of murder rates before and after states had adopted capital punishment, and (2) comparisons

Stages, developmental synchrony-asynchrony, equilibration

between states with and without capital punishment. For sorne people one type of

Severa! researchers working in a structural tradition have described sequential

1) was combined with results supporting -their original

changes in the organization of thought. The sequences generally consist of stages

position on capital punishment, and the other method (method 2) was combined

or levels meant to demarcate qualitative changes. lt is proposed that, within a

with results contradictory to their position. For other participants in the experi­

sequence, movement from one stage or leve! to the next entails reorganization of

ment method 1 was combined with results in contradiction with their position on

a forro of thought into a new form. The concept of a sequence of stages or levels,

capital punishment, and method 2 was combined with results supporting their

which is part of Piaget' s formulations and i s used by severa! others (e. g. , Damon,

method (say, method

position. The findings of the experiment showed clear cases of the assimilation

1977; Furth, 1978; Kohlberg, 1969; Langer, 1980; Selman, 1976; Turiel, 1978a; & Volpe, 1978), has recently been a tapie of controversy. Given the

of evidence to the preexisting positions. Participants in the experiment consis­

Youniss

tently found the "data" supportive of their position much more convincing than

controversies over the stage concept, it may be usefui, at this point, to describe

the disconfirming data. Regardless of how the methods and results were paired,

its characteristics.

16

The development of social knowledge

Criticisms have been Jeveled at the stage concept by sorne whose own theoret­ ical predilections are diametrically opposed to a structural viewpoint, such as those maintaining associationistic (e . g . , Brainerd, 1978), learning-behavioristic (e. g . , Bandura, 1977), or personality-trait (e. g . , Kurtines

&

Greif, 1974) posi­

tions. It would appear that those criticisms are aimed at striking at the heart of

u: 1' •1:

2 . Structuralists have proposed that thinking forms systems of organization . Critics conclude that the structural claim is that there is general consistency or unity in the individual' s thinking at each stage of development . They maintain that if the claim were correct, then thinking should be characterized by homoge­ neity. In that case, an individual 's stage should be apparent across most tasks and

structural theory in order to demonstrate its lack of validity, beca use the critics

situations. Further, it is maintained that the research findings show that thinking

make the (somewhat misleading) assumption that the stage concept is the most

is not the same across tasks and situations and that, therefore, the stage hypothe­

fundamental aspect of the theory. Criticisms have also come from sorne (e. g . ,

sis is not supported by the evidence.

Aavell, 1982; Rest, in press) whose orientation is at least partly compatible with

3. Structuralists have proposed that developmental changes in organized sys­

a structural viewpoint but who have disagreements with the stage concept and

tems of thought involve a process of interna/ regulation and equilibration. Critics

maintain that the concept is not supported by the empirical evidence.

conclude that the structuralist claim is that stage transitions are abrupt or salta­

There is consistency on certain points in the way the different critics have

tory , rather than gradual; when a new cognitive operation is acquired, it will be

characterized the presumed ·structural-developmental use of the stage notion .

rapidly applied to a large variety of task s . The critics maintain that if the claim

Those characterizations are, of course, the bases for the critica] analyses. How­

were correct , then transitions would be observed as abrupt shifts with synchro­

ever, these characterizations are not always accurate . There are important incon­

nous emergence of the new stage in a variety of tasks and conceptual domains. It

sistencies between

is maintained that the research evidence shows that changes are gradual and

use of the

(1) the way critics have characterized the structural-developmental stage notion, and (2) the structural-developmental characterization and

asynchronous.

application of the stage notion - at least, as I interpret it. Comparisons between

Have the critics adequately characterized the structural-developmental posi­

the critics' interpretations-of structural sequences with how they have been used

tions? The validity of their conclusions regarding the empirical evidence is

by its proponents (especially Piaget) would serve to clarify principies of devel­

contingent on the adequacy of the interpretations of structural propositions (if

opment and their use in the analyses presented in subsequent chapters.

the reading of the empírica! evidence is correct, which is still another matter) .

We can begin with a consideration of the features critics have attributed to

If we consider a more comprehensive version of the elements of structural

stage-sequence formulations that they have regarded as inadequate or empiri­

theory, including sorne of the features aiready discussed, it becomes evident

cally unjustified. The features to be considered come not from two or three

that the proposed empírica! tests are based on a less-than-accurate accoun t of

specific critics but represent a compilation of points made by severa! commenta­

the theory .

tors. Typically, their overriding assumptions are that structural theorists are

'' ·

17

Structure and development

l . Experience . Although structural theorists have proposed that thought or

proposing that developmental stages are , in sorne sense, autonomous, self-contained

knowledge are not copies of reality or experience, they have also maintained, as

units manifested independently of the subject's prior experiences, regardless of

detailed earlier, that development stems from the child's interactions with the

the type of task involved and across domains of knowledge. Thls general

environment, that is, from experience. Portraying stages as autonomous units

assumption, it will be argued , is an incomplete and basically inaccurate charac­

whose emergence is independent of prior experiences is more in accord with

terization of structural theory. Yet, the assumption is derived from sorne of the

innatist or maturational structurai · theories (e . g . , Chomsky , 1980) than with

theoretical parameters of the structural position. Three aspects of structural

interactional-constructivist structural theories. The interactional proposition is

theory have led the critics to propose theoretical implications and concomitant

that knowledge is formed through the subject's actions upon events, tasks, and

empirical tests:

problems , as well as through reflections upon actions (Piaget, 1976, 198Gb). It

l . Structuralists have proposed that thinking is not a copy or internalization

follows, therefore, that the child's experiences would influence the type of

&

Turiel, 1978; Turiel, 1975). The

of environmental content. Critics conclude that the structuralist claim is that

thought or knowledge constructed (Nucci

cognitive development is minimally influenced by experience. They maintain

nature of events or tasks, as well as the child's familiarity with them (the extent upon

that i f the claim were correct, then the same stages of development would be

to which the child has acted and reflected

observed regardless of the child's experiences. In tum, the structuralist position

type and leve] of thinking. However, there is an interaction between experiential

g

them) should have a bearing u pon

is faulted for payin insufficient attention to environmental influences, especially

variables and stage of development. Similar experience would be expected to

the social environment, on the child's development.

influence children at different stages in different ways.

18

The development of social knowledge

Most important, the interactional-structural model implies that there are sys­ tematic relations between the nature of environmental events ex.perienced and development. Fundamental differences exist in the types of events experienced by young children, and these differences contribute to the construction of distinct domains of knowledge (Turiel, 1979, in press). As an example, it has been proposed (Piaget, 1 970b, 1976) that a distinction ex.ists between experiences of a physical-empirical nature, which are closely related to the development of physi­ cal concepts, and logical-mathematical experiences entailing the coordination of actions, which are closely related to the development of 1ogical-mathematical cognition.

,..

and development

19

Icnowledge (based o n cognitive operations) and inforroational knowledge derived from the figurative activities of imagery (Piaget & Inhelder, 1971), imitation (Piaget, 1951), perception (Piaget, 1969), and memory (Piaget & Inhelder, 1 972) . From a structural viewpoint, therefore , there is both homogeneity and hetero­ geneity in cognition. The homogeneity expected from a stage or leve! of devel­ opment is restricted to thinking within nartowly defined conceptual domains. Variations in an individual's thinking would stem from domain differences and from the intersections of inforroational and conceptual knowledge.

The relation of experience and social cognitive development is dealt with in

3 . Developmental synchrony or asynchrony . The proposition that thought is organized within domains is inconsistent with the idea that developmental transi­

sorne detai1 in this vo1ume. Analyses of how different types of social events and

tions reflect abrupt changes across tasks and domains (general developmental

interactions are related to the-development of distinct domains of knowledge is

synchrony). It is proposed, instead, that developmental transitions, which are

considered in Chapter 3 . The relation of social judgments to prior familiarity

hypothesized to be internally regulated (equilibration), occur within domains.

with tasks and issues is considered in Chapter 4.

t::

Structure

Again, at least two propositions can be made regarding the nature of such

2. Homogeneity and heterogeneity of thought. The structural proposition that

within-domain transitions. One is that transitions are abrupt, involving an imme­

thinking forros organized systems does imply consistency of thinking across tasks and situations. However, it is not necessari1y the case that the stage notion

diate shift from one stage to the next. An altemative proposition is that transition entails a dual process of self-correction of one forro of thinking and the emer­

implies that all forros of thinking are part of the same system of organization. At

gence of a new form and is, therefore, gradual.

leaSt two alternative propositions can be made regarding the boundaries of sys­ tems of organization. It may be proposed that thought is organized in such a way

The interpretation of the developmental process in concepts of social conven­ tion (see Chapter 6) is that transitions include a phase of negation of a system of

that it encompasses all areas, al! tasks and-domains, that it encompasses the mind

thought, through criticalness of its structure, followed by the construction and

as a whole. Or it may be proposed that thought is organized within narrower

affmnation of a more advanced set of concepts. This interpretation is consistent

boundaries and in accordance with domains. As stated in Chapter l , one of the

with the cross-sectional and longitudinal findings. It is also consisten! with

aims of this volume is to explícate a domain-specific model of the development

longitudinal and experimental findings in research on moral judgments (Damon,

of social concepts . The proposition is that social knowledge is organized within

1980; Turiel , 1974, 1977) and on nonsocial cognitive development in childhood

domains (the moral, societal, and psychological) and not across domains. Addi­ tional sources of heterogeneity in thought stem from the differences between

(lnhelder & Sinclair, 1969; lnhelder, Sinclair, & Bovet, 1974; Strauss, 1972) and adolescence (Langer, in press). Indeed, a consensus can be said to exist

social-inforroational knowledge and social-conceptual knowledge.

among all these researchers that developmental transitions are gradual.

What amounts to domain distinctions have already been mentioned with re­ gard to nonsocial cognitive development, as we11. In his earliest work Piaget

Moreover, all these findings are consistent with the principie of equilibration (Langer, 1969, in press; Piaget, 1977, 198Gb; Turiel, 1969, 1974). The hypothe­

( 1 923, 1929, 1932) did hypothesize structural relations among varied aspects of

sis is that the process of change from one stage to the next is regulated by

thought (this issue is considered in Chapter 7). However, this hypothesis, which

equilibration: Development is continually directed toward increasing equilibrium,

may imp1y that a stage encompasses wide boundaries , was subsequently dis­ placed by Piaget. His conception of genetic epistemology is based on a concern

so that each stage is a more equilibrated state than the previous one. Equilibrium, however, does not simply mean adjustmen t or conforroity to externa! pressures. There are two interrelated aspects to structural equilibrium. One refers to the

with the epistemology of domains of knowledge in analyses of cognitive devel-. opment. In Piaget's research, the development of logical-mathematical knowl­

equilibrium or coherence of a system of thinking. The second refers to an

edge is distinguished from concepts about the physical world. We have already seen that this kind of a distinction is evidenced in infancy through the differences between means-ends and part-whole transformations (Langer, 1980). Further

effective way. That is, each stage of development represents a more equilibrated means of understanding the environment than the previous stage. When new

and not insignificant distinctions are made between conceptua1-transformational

inforroation, evidence, or propositions are assimilated, the coherence of cogni-

understanding of the environment in the most powerful, comprehensive, and

20

The development of social knowledge

tive organization is maintained . It is disequilibrium, which is characterized by

therefore, that interactions with fundamentally different types of objects and

conceptions of inadequacies, contradictions, and inconsistencies in the existing

events should result in the formation of distinct concepts.

way of thinking, that can result in activities producing new information leading

From the perspective of partial structures, therefore, stages or levels of devel­

to the reorganization of thought. If the child's existing way of thinking inade­

opment

quately handles encountered events , the result may be disequilibrium that can

situations. The proposition is that there is neither a general structure of mind as a

precipitate a gradual process of reorganization into a form of thought that more

whole to be identified nor are there so many domains of knowledge that we are

are not

autonomous, self-contained units manifested across tasks and

adequately coordinates the child's previous conceptions with the novel, discrep­

left with a series of elements but no systems of organization. A prograrn of

ant events .

research, th,erefore, requires identification and definition of the boundaries of

Therefore , a structural state constitutes the basis for environmental interac­

conceptual domains.

tions that may lead to inconsistencies and contradictions. As has been shown by

The analyses presented in this volume deal with domains in social knowledge,

longitudinal research with subjects identified as undergoing transition (Turiel,

on the assumption that social concepts are not all of one kind. Severa! interre­

1974, 1 977), change involves conflict over two types of concepts , those of the

lated tasks are involved. The first task is to identify and define the domains under

3 morality and social convention are defined further

prior stage and those of the emerging stage. Put in this form, transitions consti­

investigation. In Chapter

tute an identifiable phase and cannot be said to occur in sudden , abrupt fashion.

and criteria are presented to delineate their boundaries. A second task i s to

The portrayal of a developing child who shifts abruptly from one stage to another

examine the experiential sources of developrnent. The domain criteria provide a

does not accord with structural analyses of conflicts, nor with the discrepancies and contradictions that

are

proposed to be essential components of the equilibra­

basis for studying the types of social experiences that stimu!ate development within each domain, as is also discussed in Chapter 3. Two other types of

tion process . 1 Nor does it accord with the frequently observed phenomenon,

investigation stem from the proposition that morality and social convention

1981, in press), of the appear­

constitute two distinct structural and developmental systems. One type is de­

labeled "U-shaped behavioral growth" (Strauss, ··'

21

Structure and development

ance of a behavior at a certain age, its disappearance sometime later, and its

signed to test if the distinction is made by subjects of different ages and to

reappearance at a still later age. Sorne phases in development that seem to reflect

determine the criteria that they use in their judgments about each domain. A

a lesser achievement than that of an earlier phase actually represent conceptual

series of such studies is discussed in Chapter

conflicts and cognitive advance.

identified, developmental analyses would focus on sequential changes in the

4. Insofar as valid domains are

organization of concepts within each domain. The development of each type of

l it�· 11 .. ,¡::;

concept is considered in Chapters

6 and 7 .

Domains of knowledge and partial structures Cognitive structures are partía! in that they encompass delimited domains of ..,

knowledge; thinking is organized within the boundaries of fundamental catego­ ries (e.g., logical-mathematical thinking, moral judgment). Development within a domain entails reorgan izations of thought, so that separate developmental sequences can be identified for each domain. One basis for the proposition that concepts are organized within domains rests upon the idea that they are con­ structed through the individual ' s interactions with the environment. Conceptual knowledge is constructed through an interactive process, but it is not causally determined by the environment. Because such constructions originate from the individual's interactions, they are influenced by the environment. It follows, 1

Piaget's fonnulations of stages of cognitive development include the specification of substages within each stage. For instance, six substages have been identified within the stage of sensorimotor development (Piaget, 19361 1963). With the inclusion of substages in each of the stages of cognitive development, Piaget proposed a gradual process of development because the shift from the start of one stage to the start of the next stage involves progress through severa! phases.

Methods of research As with any program of research, various methods have been used in studies of structural development. These have included observational studies, controlled experiments , and a type of interview labeled the clinical method (Piaget,

1929).

The primary methods used in the studies of social cognitive development dis­ cussed in this volume were systematic observations and variations of the clinical interview method. Sorne of the studies were conducted in the everyday or natu­ ralistic contexts of the children's lives, such as schools or playgrounds, and others were conducted in the context of the "laboratory . " In this case, laboratory simply means that the researcher administered a given procedure to the subject in a separate room (e . g . , in the child's school) set apart at the time for the purpose of conducting the research. The use of the laboratory or the naturalistic context depended on the purposes ofthe research and the questions addressed. For instance, studies of the relations

22

The development of social knowledge

ated theoretical frameworks through the use of what is referred to as psychomet­

Chapter 3) have been conducted in naturalistic contexts in which the social

ric criteria (Campbel!, 1960; Cronbach & Meehl, 1955). A basic difference between the two methods of assessment is that clinical interviews are aimed at describing the organization of thought and, therefore, are

conducted in the laboratory context so as to enable clase and detailed gathering of data. One of the potential disadvantages of the laboratory context is that the

1969; Wemer, 1937), whereas psychometric measures are designed to elicit

with those encountered in naturalistic contexts. Thís means that it is necessary to

responses reflecting a pass or fail on test items that can be scored mechanically, without any interpretation on the part of the scorer. Consequently, it must be asked: If one begins with the theoretical assumptions that there are underlying

discussed in Chapter 4.

structures or organizations of thought that are related in something other than a

It should be noted that there is a sense in which the distinction between natural and laboratory contexts is artificial - depending on the topic of the investiga­

one-to-one correspondence to answers to problems or conclusions on tasks (e. g . , of the pass-fai1 variety), then i s it methodologically reasonable to replace the

tions. In research on the structure of thought the distinction is not val id, because

interview method with psychometric tests or to apply psychometric criteria in assessments of validity? The answer to this question, it will be argued, is No! This

reasoning about problems or events occurs natural! y in and out of the context of oogoing social interactions. The type of reflective thinking that occurs in the

does not mean that there are no restrictions on the standardization or objectivity of

research laboratory setting is no less real or natural than reflective thinking that

the clínica! method or that there are no criteria for assessment of validity. Quite to the contrary, standardization and objectivity have always been motivating goals of

occurs in many nonresearch settings? Specific research methods are detailed in the context of the discussions, in Jater chapters , of particular studies. The remainder of this chapter is devoted to discussion of the rationale and features of interview methods used in research on structura1 development. The reasons for amplifying on the clinica1 interview are

•.

designed to obtain data on processes of reasoning and not just on conclusions or products (Kohlberg, 1969; Langer, 1969; Luria, 1976; Piaget, 1928, 1929; Turiel,

subject's familiarity with the events or stimuli presented may not be as great as control for the familiarity of stimuli through procedures that provide the re­ searcher with its independent assessment, as was done in sorne of the studies

1' ;

23

of children's social interactions to domains of social knowledge (as discussed in interactions occur (e. g . , preschools, elementary school classrooms , and play­ grounds). By contrast, most of the studies of children's social reasoning were

,,

Structure and development

twofold. First is its significance for structural-developmental research and, espe­ cial! y , its widespread use in studies of social cognition. The second reason is to clarify the aims and uses of the method, especially with regard to the differences between the clinical interview and so-called standardized testing in the psycho­

the users of the clinical interview, and there are criteria to assess the validity of the methods and the related theory. However, the criteria are somewhat different from psychometric criteria and closer to general scientific criteria of hypothesis testing and theory validation. Accordingly, a brief description of the clínica! method i s provided (for more extensive discussion see Cowan, 1 978; Damon, 1977; Piaget, 1929), followed by discussion of the reasons why tests with mechanical scoring

methods are not appropriate replacements of the clínica! method and why psy­ chometric criteria

are

inadequate for evaluation of structural theories.

metric tradition. ''!,

A major question regarding assessment and scoring in the clinical method pertains to the standardization of interviewing and the objectivity of response scoring. An incorrectly presumed lack of standardization of assessments and objectivity of scoring has sometimes led to attempts to translate the procedures into "tests" that do not require the interviewing of subjects and that are scored mechanically. In addition, confusions about method and theory are evident in efforts to evaluate (e.g., Kurtines & Greif, 1974) the clinical method and associ2 The methods of assessing reasoning in natural and laboratory contexts are not in opposition and

should be seen as complementary. There are methodological advantages and disadvantages to each method. The laboratory methods allow for specification and control of variables, as well as experi­ mental manipulations to pursue predetermined questions or hypotheses. The laboratory methods also provide data on reflective thinking, which can then be used in the study of thinking in naturalistic scttings. Although naturalistic settings are not often amenable to the control of variables, they do provide stimulus events· that are not readily reproducible in the laboratory. Naturalistic seuings also provide data on thinking in the situational conteXt, which can then be used in the study of reflective thinking in laboratory seuings.

The interview method The process of formulating methods of research cannot be dissociated from the topics of investigation or the type of information that one intends to uncover. To be productive, the investigator's methodologies must be related to hypotheses and theoretical propositions. If one begins with a set of propositions requi.ring exploration but fails to use methods that are adequately adapted to the task or that are insufficiently powerful to tap the proposed phenomena, then the proposed phenomena will, essentially, go uninvestigated and related hypotheses will re­ main untested. As an obvious example, the proposition that an understanding of cognition is necessary for explanations of the psychology of human functioning is incompatible with radical behavioristic methodology of the sort that would be based on the assumption that cognition is irrelevant to psychological investiga­ tion (Watson, 1924).

24

The development of social knowledge

Structural-developmental theorists have proposed, as already discussed, that thinking is most adequately explained by describing its organizing principies, that its development is not a continuous accumulation of content, that develop­ ment involves qualitative changes forming a systematic sequence, and that phases of transition entail conflict and contradiction. The clinical interview is one of the methods originally designed (Piaget, 192711960, 1928, 1929) and subsequently refined (e. g . , Inhelder & Piaget, 1 95 8 , 1964; Piaget, 1952) specifically to pro­

and the aims of the investigation.

A feature of the clinical method is that the interviewer decides during the

course of the interview which probing questions to ask. In an adequately con­

structed interview with a well-trained interviewer the probes are also systemati­

about particular issues and not just what he thinks or how correctly he salves

response given by the subject. That is, one type of response could require one set

analyzing products are the repeated observations of U-shaped behavioral growth (Strauss, in press). According to Strauss, the phenomenon of a disappearance of

..

structed interview the questions are not random or unrelated to proposed hypotheses

cally determined. One type of probe is straightforward and predetermined in the

problems. Research findings confirm the need for methodologies that examine process in addition to products. It has been shown that children's solutions on a task or problem or their correct or incorrect answers to questions are not complete indicators of type or leve\ of conceptualization (Karmiloff-Smith & Inhelder, 1975; Kohlberg, 1963; Piaget, 1928; Sugarman, 1979; Turiel , 1969, 1974, 1977). Perhaps the clearest and most striking demonstrat ions of the inadequacy of solely

,. 1

and the associated questions should each be carefully designed to yield informa­ tion relevant to specific issues addressed by the researcher. In a properly con­

vide a means for eliciting data that would inform these structural propositions. In

particular, the clinical method is designed to ascertain how an individual thinks

.,. , ..

25

Structure and development

an earlier behavior, which actually reflects an advance in underlying perfor­ mance, has been found in severa! areas, including language acquisition, artistic expression, conservation of quantities, and motor ski lis. Another example comes from the experimental work of Kanniloff-Smith and Inhelder ( 1975). They observed that younger children (4V2- to 5 V2-year-olds) appeared to be more

successful in solving a balancing-blocks problem than were older children (5V 2to ? Y2-year-olds). Analyses of the children's strategies revealed that the youn­

ger children were using a trial-and-error procedure that was simpler than the hypothesis-testing procedure of th� older children (which led them to the incor­ rect solution).

As a means of obtaining information relevant to analyses of processes of thought, the clinical method consists of a set of tasks pertaining to a domain (such as number, classification, seriation, or morality) and a closely associated interview, which in eludes a series of predetermined questions and probes based on specific hypotheses. In sorne cases, the subject is presented with a task of the sort that has objects and transformations of objects to observe· and/or manipulare. The subject is instructed to both sol ve the task and answer questions that pertain to the ways the task-related objects and transformations are conceptualized and to the mode of reasoning involved in solving the task. In sorne cases, particularly in

research on social cognition, verbally presented problems (e. g . , in story form)

are solved and the questions pertain to the conceptualization of the elements of the problem and mode of reasoning. Concrete tasks, verbally posed problems,

interview schedule. Altemative sets of probing questions are contingent on the of probes, whereas a different response would require a different set of probes (or perhaps no pro bes). 3

Other types of probes are not always predetermined in the interview schedule but represent questions introduced in the course of the interview to obtain ade­ quate responses relative to the objectives of the investigation Such probes are .

used for two interrelated reasons. One is that a question on the interview sched­

ule may be misinterpreted by the subject. Probes serve to clarify questions, so that the subject understands the question as it is meant by the researcher and responds accordingly. The other reason for probing questions stems from the

assumption that subjects interpret questions and do not merely react to them as stimuli. Remember that central to structural theory is the proposition that the

stimulus-response relation is reciproca): The individual interprets stimuli on the

basis of type or level of thinking. Probes are, therefore, deemed necessary in

order to determine how the subject is interpreting the task and the questions

.

Probing questions of a systematic nature are one of the ways to control for the

different meanings that can

be attributed to the same stimuli

.

It should be apparent that the administration of a clinical interview is con­

ducted in accordance with strict guidelines. Therefore , interviewers must be well-trained in severa! respects. An interviewer must be well-versed in structural

theory and in the specific hypotheses and aims of the particular interview admin­ istered . Furthermore, the interview er must be adept at stimulating the subject to

deal with the task or problem presented and to respond to questions. lnterviews that produce more information from the subject yield richer and more reliable

data. Indeed, a well-trained interviewer is one who comprehends the experimen­ tal nature of the method. As Piaget maintained ( 1 929, p. 8), "The clinical method is experimental in the sense that the practitioner sets himself a problem, 3 A useful technique in clinical interviews is what Piaget tenned "countersuggestions." In a countersuggestion the subject is presented with a judgment differing from the one he has giveo (in the fonn of, for example, "another child told me that . . . "). Countersuggestions test the limits of the certainty with which the judgment is maintained and provide a stimulus for furtlier exploratio? of reasoning. In some of the studies discussed in Chapters 3 , 4, and 5 the tdea of countersuggestlons was translated and expanded into systematic procedures for obtaining data on specific issues.

26

The development of social knowledge

mak:es hypotheses, adapts the conditions to them and finally controls each hy­ pothesis by testing it against the reactiorís he stimulates in conversation." Just as the theoretical premises and research aims are related to the methods used in data gathering, so too are they closely related to methods of data interpre­ tation. Because the aim of gathering data through the clínica! method is to obtain information about the organization of thought and sequential transformations, it follows that methods for coding responses would include descriptions of such systems and a basis for distinguishing them sequentially. Therefore, procedures for coding responses are aimed at describing, in as detailed and precise a fashion as possible, the relevant elements of thought and their interrelations. A coding procedure attempts to characterize thought through criteria for classifying re­ sponses and descriptive categories that can be used by trained coders. The coding procedures are part of the theoretical framework in that the categories are de­ scriptions of organizations of thinking. Coders, too, must be well trained so as to understand the theoretical framework and the technical details of the categories; to be able to interpret the bases for discriminating between responses that are and are not germane to the domain of the investigation;4 and to be able to objectively interpret responses and apply the coding system.

The clinical method and psychometric measures compared

:.::.:

�1

We can now retum to the standardization and objectivity of the clínica! method as compared with psychometric modes of assessment. Because the clínica! inter­ view allows for probing questions, whose use is dependent on the subject's responses and which may vary somewhat from one subject to another, it has been claimed that it lacks standardization (e.g., Enright, Franklin, & Manheim, 1980; Kurtines & Greif, 1974). Because the scoring procedures are not mechanical and require an understanding of the coding system as well as judgment in its applica­ tion, it has been claimed that they are not objective (Kurtines & Greif, 1974). In turn, it is claimed that psychometric measures are more precise because of their standardization (i.e . , the same questions are administered in exactly the same way to all subjects) and objectivity of scoring ( i . e . , scoring is mechanical , requiring no interpretation on the part of the scorer). To evaluate these claims it would be useful first to consider a concrete exam­ ple of a psychometric measure. Psychometric measures of intelligence provide a good example, because the clinical method was originated by Piaget ( 1 928, 1929) as a way of ameliorating, given his theoretical orientation, what he re­ In clínica! interviews a subject may provide responses that do not correspond to any of the coding categories. Responses that cannot be scored are only a problem if the coding procedures do not have clear criteria for distinguishing responses relevan! to the domain of judgment from those that are not. Coding procedures with such criteria obviate the necessity for coding aU responses, when sorne of them are not gerrnane to the topic under investigation.

•

27

Structure and development

garded as their methodological shortcomings. Typica!ly, psychometric measures ·0f intelligence contain a series of items (problems, tasks, and questions) de­ signed so that the responses can be recorded as correct or incorrect. A tally of the number of correct responses produces a score that constitutes the basis for com­ parison with standardized norms. The tests are calibrated to age norms in that items vary according to their difficulty. However, test items are not calibrated to type of intelligence or knowledge. Intelligence tests, such as the Stanford-Binet (Terman & Merrill, 1937), contain items pertaining to a variety of areas, includ­ ing number, logical inferences, classifications, vocabuhizr , information (e. g . , naming the days o f the week), and motor skills. All items are scored on a pass-fail dimension and are weighted equally. The intelligence-testing' tradition, too, illustrates that methods of study are closely related to a set of theoretical propositions. In this tradition intelligence is defined as a trait or capacity possessed by individuals. Therefore, intelligence is defined quantitatively; it is an amount of the trait or capacity possessed by an individual, relative to others in comparable groups (for sorne theorists the amount can shift with experience, but for others it is primarily fixed by inherited factors). Insofar as additional theoretical issues regarding intelligence are considered, it is through proposed relations of the capacity to other variables (what Cronbach & Meehl, 1955, have termed the "nomological network"). In itself, intelligence is operationalized as the capacity measured by inte1ligence tests. The contrasting structural position is that "intelligence" is a topic of direct investigation in the formulation of theories of thought and knowledge. The study of intelligence is the study of cognitive development. In the study of cognitive development, the sorts of logical, mathernatical, and physical tasks found in intelligence tests would be presented to children through the clinical method in order to investigate their forms of reasoning. This entails obtaining more than solutions to the problems, as is well illustrated by Piaget's (1928) adaptation, in his early work, of items from the Binet-Simon ( 1916) tests of intelligence. For instance, the Binet-Simon test included a set of "absurd sentence" items of the following sort (as cited in Piaget, 1 928, p. 63): (a) (b) (e)

A poor cyclist had his head smashed and died on the spot; he was taken to hospital and it is feared that he will

not

recover.

1 have three brothers: Paul, Emest and myself. Someone said: If ever 1 kili myself from despair l

won 't choose a Friday ,

because Friday is a bad day and would bring me ill luck.

In the Binet-Simon version, the subject's statement of what is absurd or wrong with each sentence is scored as correct or incorrect. Piaget used these same items to study elements of thought by examining how children understand the logical contradictions and reciproca! relations (e . g . , in the brothers example) in the sentences. By explaining how children reason about these items, Piaget focused

The development oj social knowledge

28

on cognitive activities like reciproca] relations, on the use of premises, deduc­

It is certainly reasonable, however, to argue that the theoretical.premises that

tions, hypothetical propositions, and on discriminations between empírica! and logical necessity. This is just one example representing the early work on devel­

]ead to the clínica! method are inadequate. However, on the assumption that structural theory may be correct, it is apparent that psychometric measures are

opment from concrete forms of inference to formal thought. This example illus­

not the only means of attaining adequate levels of standardization and objectiv­

trates the difference between the purposes of the clinical and psychometric methods.

ity. In fact, from the structural perspective psychometric measures are sources of

Psychometric methods are designed to obtain test scores reflective of a quantita­

inaccuracy and are lacking in controls in that they fail to account for the subject's

tive assessment of a capacity or trait, whereas clinical methods are aimed at

interpretations, for differences between the content of responses and underlying

gathering data for analyses of qualitative and developing characteristics of thought

structure, for varying means of arriving at correct responses and making errors,

processes.

and for different types of knowledge. Moreover, subjects are susceptible to

·

The example also highlights that the two approaches lead to differing strate­

As can be seen, both standardization and objectivity are goals of the clinical

over, the methods of each approach are judged as lacking in standardization and

method - as they should be. In the clinical method, however, standardization is

objectivity, if they are viewed through the lens of the opposing theoretical

sought through specification of the parameters of the stimuli (task, question) and

perspective. Therefore, the perceived degree of standardization is not a purely

the subject's interpretations of stimuli. lt is presumed that the goal of standard­

In the psychometric approach standardization is achieved by eliminating or

1· ,..

ization is furthered by extensive questioning and probing in a given interview. The adequacy of standardization can be assessed in two ways. One is through

drastically reducing the role of the tester. The uniformity of the assessment

measures of test-retest consistency over short periods of time. lt has been dem­

instrument is considered free of bias as long as the items are presented in exactly

onstrated that the clinical method can have high test-retest reliability (Damon,

the same way to all subjects on a sheet of paper (to a subject who reads them)

1980; Colby et al., in press; Selman, 1980). The second way of assessing

or in the tester's words (to a subject who hears the�). The elicited response is

standardization is through the replicability of findings. If a measure yields

simply recorded by the subject (if in written form) or by the tester (if in oral

similar results in studies by different researchers, it can be concluded that

form). Insofar as a tester deviates from the questions to elicit other responses

the measure is stable. Studies on social cognitive development based on the

(e.g . •. through probes), it is assumed that bias is brought into the study (Enright

clinical method have shown high replicability (e.g., Damon, 1977; Kohlberg,

& Greif,

1974). This realist conception o f standardization

1976; Selman, 1980; and the studies reported in this volume). The replicability

follows from ( 1 ) the assumption that the subject's response to stimuli is unilat­

of research on nonsocial cognitive development by Piaget and others has, as put

eral and not interpretative, and (2) an attempt to assess a quantitative dimension

by Cowan ( 1 978, p.

of a phenomenon (e. g . , intelligence) that is presumably tapped at the leve! of

field."

et al. , 1980; Kurtines

.,.

momentary influences because their responses are.left unmonitored.

gies regarding standardization of assessment and objectivity of scoring. More­

methodological issue but is related to theoretical perspectives .

,•'

29

Structure and development

64),

"yielded sorne of the most replicable results in the

Correspondingly, objectivity of scoring procedures is a goal of the clinical

content. The realist conception of standardization, however, does not follow from the

method. However, objectivity is not regarded as synonymous with mechanical

structural-developmental theoretical propositions. Therefore, it does not follow

methods of scoring. If it is proposed that thinking forms organized systems, then

that users of the clinical method should accommodate to the psychometric con­

it follows that its adequate characterization would require nonmechanical, com­

ception of standardization. At the simplest leve! the argument can be put as

plex, and interpretative forms of coding. It is for this reason that coding proce­

follows: If one propases that a stimulus is interpreted by the subject and that the

dures need to consist of precise!y defined systems of classification and categories

interpretations may vary among subjects in systematic ways, then the methods

of judgment. In this context, it is neither plausible nor desirable to eliminate

to be adequate to the task, will account for the subject's

scorer judgment. A scoring procedure is sufficiently objective if it can be used

interpretations of the stimulus and not solely for the subject's noninterpretative

with consistency by independent (but trained) coders. There is no question that

reaction to it. This is a simple, logical point! Accordingly, it is not reasonable to

coding procedures developed through the clinical method can be objective, as

claim that the method is inadequate because it allows for the subject's interpreta­

evidenced by the high interrater reliabilities reported by severa! researchers (e. g.,

tion of the stimulus or that it should be replaced by a method that solely records

Broughton, I978a; Damon, 1980; Kohlberg, 1969; Selman, 1980; Turiel, 1966;

and tailies responses to stimuli.

and as described in Chapter 6 of this volume).

used, if they

are

30

The development of social knowledge

Evaluation of structural theory by psychometric criteria

• .,

1

'

Misunderstandings of the aims and procedures of the clinical method, therefore, have resulted in misconceptions about standardization and objectivity. These misconceptions have been further compounded by translation of sorne general propositions of structural theory into psychometric criteria, with resulting mis­ conceptions of both the 'theoretical propositions and the ·research findings. In addition to test standardization and mechanical scoring requirements, psycho­ metric criteria include predictive validity and construct validity (Cronbach & Meehl, 1955). These criteria, as used psychometrically, are specific to the vali­ dation of tests. As stated by Carnpbell ( 1960, p. 546), "Test validity and test reliability are not concepts belonging to the philosophy of science. Instead they are concepts which have developed in the course of mutual criticisms of test constructors and test users, concepts which relate to the implicit and explicit claims of test constructors and test salesmen . " Psychometric criteria were designed to provide ways of assessing the scientific status of tests and related theoretical propositions. The importance of evaluating any empirically based theoretical framework and the necessity of testing hypoth­ eses is, of course, consensually accepted. However, the special terms of psy­ chometric criteria, like predictive and construct validity, are specifically germane to the evaluation of psychological tests. Most important, the overriding feature of the rules of psychometric validation distinguishes it from the usual norms of evidence for assessing the adequacy of a theoretical framework: With psycho­ metric criteria, the test stands at the core, as a source of quantitative assessment, constituting the standard to which a set of propositions (the nomological net­ work) relate. The test scores do not represent descriptive or explanatory proposi­ tions. In structural theory, by contrast, descriptions of stages or levels of development are in themselves theoretical propositions designed to explain phenomena, such as logical inferences or moral judgments. The descriptions of the organization or types of thinking constitutes one element of a theory that may be related to other elements through additional theoretical explanations. As examples, theoretical formulations may explain relations between types of thinking and behavioral decisions, or the ways in which individuals pass through a developmental se­ quence. In such a case, a comprehensive theory would include explanations of thought processes, of behavioral decisions, of movement in a sequence, as well as the interrelations among these elements. It is important, therefore , to distinguish between test validation, on the one hand, and the use of empirica!'evidence, on the other. Nevertheless, psychomet­ ric criteria have been applied to structural theories by those who characterize a clinical method and related structural descriptions simply as a psychological test. One of the most extensive and perhaps most influential efforts at applying these

Structure and development

31

psychometric criteria to structural-developmental research was made b y Kurtines and Greif ( 1 974) in their purported evaluation of Kohlberg's ( 1 963, 1969) theory and research on moral judgments. 5 The many misconceptions and confusions in the Kurtines and Greif evaluation have been carefully and insightfully discussed elsewhere (Broughton, 1975, 1978b). For the present purposes , their analyses provide concrete examples of how the notions of predictive and construct va1idity are misapplied to structural theory. The notion of predictive validity reflects a behavioristic bias in that one of the requirements of a test is that it be predictive of behavior. This is because psycho­ logical tests are usually designed to provide consumers with a means to predict performance. Many intelligence tests are designed for use in educational settings to predict academic performance. Personnel-selection tests are designed to make predictions about occupational success. This model was adopted by Kurtines and Greif in their assertion that the validity of Kohlberg's methods of assessing moral judgments depended, in part, on the extent to which they were predictive of moral behavior. Moreover, a rather stringent test of predictability was imposed. A one-to-one correspondence between stages of moral judgment and behavior was expected, so as to demonstrate the discriminative validity of the measure. However, like other structural analyses of social cognition, Kohlberg's methods were designed not to predict behavior but to describe judgments. The judgment­ action issue is one that requires explanation . The explanation of relations be­ tween thought and action, and especially their developmental relations, is a problem that can be approached through a range of hypotheses. At one pole, it would be possible to hypothesize that moral judgment and moral behavior are largely unrelated. For instance, it may be proposed that moral judgments serve real and significant functions separate frorn those served by behavioral systems. Whatever the proposed reasons for the d iscontinuity, if the hypothesis were correct, then valid descriptions of judgment would not be predictive of behavior. Determining the relations of thought and action is a complex and difficult problem in psychological theory that is not reducible to prediction or correlation. Chapter 9 considers social behaviors and their relations to domains of social judgments . It is proposed that moral judgment is only one componen! in the process of coordinating different domains of judgment in behavioral decisions. In such a case, the absence of a strong empírica! association between moral judgment and behavior would not indicate a lack of validity of the assessment of moral judgment. � Kurtines and Greif mistakenly assumed that Kohlberg' s method was a projective test. The differ­

ences between projective methods and the clinical interview are unambiguous. Projective tests are highly unstructured stimuli; responses to them are presumed to reflect personality characteristics and unconscious dynamics largely inaccessible through direct methods . The clinical interview is struc­ tured, task-oriented, and presents thc subject with concrete problems to solve.

The development of sociál knowledge

32

Similar! y, the criterion of construct validity fails to account for explanations of the relations among different variables in a theoretical framework. Construct validation is meant for test scores that are not descriptive of an underlying construct (Cronbach

&

Meehl, 1955, p. 282) but that are used in relation to

scores on other measures. Therefore, construct validation is also a predictive

3

Social experience and social knowledge

notion. It is used to assess the adequacy of test scores in predicting other vari­ ables or other test scores . To evaluate the construct validity of a structural assessment it is necessary to transform it into a test score that supposedly predicts other variables. In an arbitrary fashion, Kurtines and Greif established , as one test of the construct validity of the assessment procedures, the proposition that the moral judgment stages described by Kohlberg form a unidirectional and invariant sequence. In other words, the validity of the hypothesis that children progres s through the stages step-by-step in the prescribed sequence and without regressions was one of the ways of determining if the descriptions of moral judgments were valid. Again, the relations among different components of a theory were misinterpreted as predictive variables. Alternative hypotheses can and have been proposed regarding the assessment of types of thought and processes of change. It may be

proposed (Wemer & Kaplan, 1963), for instance, that regressive changes are

necessary for progressive changes within a sequence. Still other developmental hypotheses are plausible within the context of a particular assessment and de­ scription of forms of thinking (cf. Flavell, 1982). The theoretical task is to explain the structures of knowledge and the processes of development. Components of a theoretical framework are not static constructs that validate or invalidate a central testing instrument. Rather, they are explana­ twns of phenomena and of the interrelations among variables. The differences between test validation and assessments of the validity of psychological theory were recognized by Campbell ( 1960) in bis rebuttal to Bechtoldt' s ( 1 959) cri­ tique of the topic of construct validity. Campbell' s counterargument was simply that Bechtoldt's critique was directed at philosophy-of-science issues and that the scope of construct validation is delimited within the narrow confines of test validation ( 1 960, p. 546): While not denying the presence of a serious philosophical disagreement nor its relevance to psychology, thi� paper will emphasiz� the c mmon ground implicit in psychology's ? tradltiOn of test vahdauon efforts. The phtlosophJcal d1sagreement will remain, but it need not produce a lack of cons nsus about desirable evidence of test validity. Bechtoldt's � . argu me nt JS mdeed more a�amst the role of construct validity in discussions of philosophy of �c1ence and psychologJcal theory, rather than an objection to spec ific statements of desuable

ev¡dence

of test validity

contained under that

rubric.

Two propositions regarding the child's interactions with the environment were presented in the previous chapter: that interactions are reciproca! and that qualita­ tively different types of events experienced by the child produce distinct domains of thought. Responses are not unilaterally determined by stimuli and, therefore, lcnowledge does not stem directly from experiences. The child gains knowledge by acting upon and abstracting frorn events. In addition to experiencing events, individuals select, interpret, and systematize elements of their experiences. How­ ever, social experiences are systematically related to the developrnent of social concepts . Just as social concepts are not al! of one kind, social experiences too are not all of one kind. There exist different forms of social experiences associ­ ated to development within domains of social knowledge. The research discussed in this chapter examines elements of social interaction as they relate to the formation of moral and social conventional concepts . Before considering these studies, it is necessary to provide working definitions of the domains under investigation and the criteria that delineate their bound­ aries. One of the definitional problems in the social domains is that similar terminology is often used to mean different things. Jn everyday discourse , the same label, including the labeling of events as moral or conventional , may be applied in different ways and with different implications. Terms that are associ­ ated with moral discourse, such as "good," "should," and "ought," are also frequently used in nonmoral senses. In a philosophical essay on morality , Harman

( 1 977) provided an informative example of how the term "should" can be used in moral and nonmoral ways. A moral use of the term might be, "The thief should

not steal . " Moral reasons would then be given to support this prescription. It

could also be stated, without endorsing the action of stealing, that "in carrying out the robbery, the thief should wear gloves." In the latter case, means-ends reasons of efficiency would be given for why the thief shquld wear gloves. Consider other examples of uses of the term "should" that can pertain to either morality or social convention. The following statement could represen! a moral use of the term: 33

34

The development of social knowledge He should not kili innocent children!

In its moral sense, the statement is

not

contingent on group membership or

societal standards . We would not say that if he wants to be part of the group, , he should not kili innocent children; nor that it would be permissible for him to kili innocent children i f he did not wish to be part of the group. By contras!, the following uses of the term are contingent on group membership or societal standards:

Social experience and social knowledge

35

relative to the societal context. In addition to the variability of conventions from one social system to another, they may be altered by consensus or general usage within a social system.

The individual's concepts of social convention are, therefore, closely related to his or her concepts of social organization. In contras! with convention, moral prescriptions are not perceived to be alterable by consensus. This is not to say that morality is fixed and unalterable . We know, for instance, that historical changes have occurred with regard to

such matters as slavery . However, the bases for those changes are not perceived as shifts in the general consensus or in social organization, but on the intrinsic merits , from the

If he wants to be a member of our group, he should wear a yellow jacket. If he wants to play baseball with us, he should play as hard as he can.

In these statements the term "should" is used in a nonmoral sense. The behav­

moral point of view, of one type of action over another. Again, in contras! with conven­ tion, in the moral domain actions are not arbitrary, and though moral prescriptions form part of social organization, they are not defined by social organization nor is their ration­ ale based on their status as implicit or explicit regulations. The individual's moral

�

attainment of the goals of the activity. It would be acceptable for him to wear a

prescriptions (e.g., regarding killing and the value of life) are de ermined by factors . _ inherent to social relationships, as opposed to a particular form of soc1al orgamzauon. An individual' s perception of an act sueh as the tak.ing of a life as a transgression is not

brown jacket if he did not wish to be part of the group. There is no requirement

contingent on the presence of a rule, but rather stems from factors intrinsic to the event

that one be part of the group or engage in the activity of playing baseball .

(e.g., from the perception of the consequences to thc victim). This means that moral

ioral requirement is directly related to group membership or participation in

issues are not perceived as relative to the societal context. The moral theories forrned by individuals are based on concepts regarding the welfare of persons, the rights of persons, and justice, in the sense of comparative treatment of individuals and means of distribu­ tion . [Turiel, in press, pp. 3&-40)

Morality and social convention Definitions of morality and social convention consisten! with the two .examples of the use of the term "should" have been presented previously (Nucci

& Turiel ,

1 978; Turiel , 1979, i n press). In reviewing those definitions, they will be ex­ panded somewhat in relating them to corresponding philosophica! conceptions. Indee.

u e .,

-�

V)

V)

�

�

B �

o. "' .. ·z .....i

expected, the moral transgressions were evaluated as more serious than the conventional transgressions. In addition, subjects stated that moral transgres­ sions were deserving of a greater amount of punishment than conventional transgressions.

"' e

o; E (/) "'

·¡:¡ u :::>

z .....i

The differences in thinking about moral and conventional transgressions among

3- and 4-year-olds were not limited to assessments of seriousness or punitive­

ness. As indicated in Table 4. 1 , subjects were also posed two criterion-dimension questions referring to rule contingency and generalizability. For the purposes of

58

The development of social knowledge

these analyses subjects were divided into two age groups: those from 2% to 3 % years old and those from 3% to 4% years old. Both groups of subjects treated the moral events as noncontingent as to rules. It was generally stated that moral transgressions would be wrong even if no rule existed; the young children fo­ cused on the intrinsic features of the actions, so that their moral judgments were not based on the presence or absence of an externa! rule. The older group of subjects al so regarded morality as generalizable in that they said that the transgressions would still be wrong in a different context. By contrast, the younger group of subjects did not demonstrate an understanding of the generalizability of moral judgments. Their responses to the generalizability questions reflected a lack of comprehension of the issue. It appears, therefore, that noncontingency is an earlier-developing dimension of moral judgment than

generalizability. Moral transgressions were considered non-rule-contingent by all the subjects, whereas the children under four years of age did not use the generalizability dimension. The findings of the Smetana study also indicated that children of 3 to 4 years of age have not yet formed stable understandings of the types of conventional transgressions presented i n the study. The findings regarding rule contingency for conventions were equivoca!, though clearly in the predicted direction. Al­ though a majority of subjects stated that the conventional acts would be all right if no rule existed, it was not a majority large enough to attain statistical significance.

Table 4.2. Number of subjects out of 16 at each grade level sorting actions as

"Wrong even in absence of rule" Grade Action

Moral Lying• Stealing Hitting Selfishness Athlete throwing gameb

independent of personal desires?).

15 16 15

Boy entering girls' bathroom Eating lunch with fingers

8

11

College

15 16

16 15

16

16

15 16

16 16 16 16 16

o

o

o

o

2

1

o

o

o

o

o

o

15 16

14

o

Chewing gum in class• 1

o

Eating in classb Talking without raising handb

Personal

1

o

2

o

Watching TV on a sunny day•

1

o

Keeping correspondence private

1

o

o

o

o

o

o

o

o

o

o

o

o

o

o

o

o

o

o

lnteracting with forbidden friend

Refusing to joín recreation groupb

indusion of "personal" events provided a means of measuring the criterion dimension of impersonality (i.e . , ls the act considered a personal decision or is it

16 16 16 16

Addressing teacher by first name

who ranged from 7 to 1 9 years of age, were also presented with a series of moral

be considered the realm of personal jurisdiction (also listed i n Table 4.2). The

5

Social convention

Boy wearing long hair

Subjects were also presented with severa! events that, in this culture, are likely to

2

Damaging borrowed propertl

However, clear-cut results for the dimensions of rule contingency and imper­ sonality were obtained with older subjects by Nucci ( 19 8 1 ) . Nucci's subjects, and conventional transgressions as stimulus events (they are listed in Table 4.2).

59

Dimensions of social judgments

Smoking at homeb

o o

Note: Eight subjects wou1d be expected to sort an actíon as "wrong even in absence of rule" by

chance; p < .O 1 or greater when fourteen or more subjects sort an actíon as "wrong even in absence of rule" using x2 goodness·of-fit test.

"Actions sorted by subects in grades 2 and 5, but not grades 8, 1 1 , or college. bActions sorted by subjects in grades 8, 1 1 , and college, but not grades 2 or 5 .

Source: Nucci ( 1 977).

Rule contingency and ·impersonality were assessed by the following proce­ dures. Each subject was presented with the entire list of acts and instructed to

are consistent with the Smetana findings, showing that across a wide age range (3 to 19 years), evaluations of moral issues are not based on the presence or absence

group together those acts considered wrong regardless of the presence or absence of a rule for each act (this is a measure of rule contingency similar to that used in

of social rules. Table 4 . 3 shows that moral acts are considered impersonal (i. e . , not solely the "person's own business"). None of the moral transgressions were

the Smetana study). After sorting the acts on the basis of rule contingency, the

classified as actions that should be the person's own business.

subject was again presented with the entire list and instructed to group together

The sorting of the conventional transgressions under each instruction demon­

those acts considered the "person's own business" and that should not be gov­

strates sorne of the salient dimensions of social-conventional thinking. On the

erned by a rule.

one hand, subjects' evaluations of conventional transgressions were contingent

The results from these two procedures are reproduced in Tables 4.2 and 4 . 3 . Table 4.2 shows that, at each o f the ages represented in the sample, only the

on the presence or absence of a rule (Table 4.2). On the other hand, conventions

moral transgressions were judged as wrong regardless of the rule. These findings

were not regarded as legitimately based on personal inclination (Table 4.3). The most plausible interpretation for the combination of rule contingency and imper-

The deveiopment of social knowiedge

60

Number of subjects out of 16 at each grade Leve/ sorting actions as "Shouid be person' s business"

Table 4 . 3 .

Moral Lying• Stealing Hitting Selfishness Athlete throwing gameb Damaging borrowed propertyb

Social convention Chewing gum in class• Addressing teacher by first name Boy cntering girlso bathroom Eating lunch with fingers

2

5

o o o o

o o o o

o o o 5

o o 1 5

Talking without raising handb

Eating in classb

8

11

College

o o o o o

o o o o .o

o o o o o

o 2 8 o o

o 3

8 o o

o o 9 o o

were in turn ranked as more wrong than the personal actions (these differences subjects at different ages. The analyses of explanations for the ranking yielded a coding system of the types of justifications, which were closely linked to the less wrong. The justifications for the ranking of moral transgressions produced reasons of justice, harrn, and rights; the ranking of conventional transgressions was justified on social-organizational grounds; the reasons given for ranking acts as least wrong were that they are of a personal nature and only affect the actor. The Nucci study was replicated by Smetana

( 1 980) with the addition of an­

events presented to subjects were put in the form of transgressions. However, not all social actions relevant to morality take the form of transgressions. Actions o f a positive nature, such a s helping others , can have moral components. Conven­ tional transgressions generally have their counterparts in the opposing act that affmns the convention (e. g . , dressing in the prescribed mode or using titles as a form of address). In the Smetana

13

16

12

12

lnteracting with forbidden friend

12 13

13

13 13

13 13

13 13

12

15

15

12

12

13 12 13

Refusing to join recreation groupb

the moral domain were ranked as more wrong than conventional ones, which

other type of stimulus event. In all the studies mentioned thus far the stimulus

Watching TV on a sunny day• Keeping correspondence private

Smoking at homeb

The ranks corresponded directly with the domain o f the act. A t all ages acts in

rankings. That is, different justifications were associated with the acts ranked

Personal

Boy wearing long hair

61

were statistically significan!). There were no differences between the rankings of

Grade Action

Dimensions of social judgments

12

12

Note: Eight subjects would be expected to sort an action as "should be person °S business" by

chance; p < .05 or greater when twe1ve or \llOre subjects sort an action as "should be persón o s business" using x2 goodness-of.fit test. •Actions sorted by subjects in grades 2 and 5 but not grades 80 1 1 o or college. bActions sorted by subjects in grades 8o 1 1 o and college o but not grades 2 or 5 . Source: Nucci ( 1 977).

(1980) study judgments about positive actions

were compared with transgressions, using the procedures from the Nucci study. The findings demonstrated that criterion judgrnents apply to both transgressions and positive actions. One other study (Weston

& Turiel , 1 980)

examined the criterion dimensions

of institutional practice, which is closely related to rule contingency, and juris­ diction or power of authority . The procedure was to present hypothetical stories about schools in which it was the policy to permit certain actions. Four actions were specified, and only two of them were classified as moral or conventional: hitting another child and undressing in the playground. (One story described a school in which children are allowed to hit and even hurt each other; the other, a school in which children are allowed to wear no clothes . ) In each story, subjects

sonality is that the status of conventions can vary in accordance with their role in

were asked to evaluate the schoo!'s policy o f permitting the act and to predict an

the social system. Within the cultural context of the subjects of the study, the

authority's ( i .e . , a teacher's) response to a child who engaged in the act.

conventional transgressions would be acceptable if not govemed by regulations,

The purpose of the procedure used in the Weston and Turiel study, therefore,

but at the same time they are legitimately considered under social jurisdiction.

was to ascertain how children relate the policies of the institution and of those i n

Unlike actions with intrinsic consequences, arbitrary acts are judged by their

authority to their own judgments o f the actions. Subjects were asked, i n a

societal context.

concrete way, about cases in which the policy of the institutional authorities

In conjunction with criterion judgments, domains are distinguished by forms

permits the actions. Subjects at al! the ages, it was found, did not accept the

of reasoning. The Nucci study provided· evidence of the relationship between

legitimacy of the policy regarding hitting, an act in the moral domain. The

domains and reasoni¡tg through an additional set of findings. Subjects were

majority

instructed to rank the list of actions according to their degree of "wrongness"

hurt one another, saying that it was wrong because this could result in harm to

(from most to less to least wrong) and to explain their reasons for the rankings.

persons. By contrast, the majority

(86%) were critica! of a school policy that would allow people to hit and (68%) at all ages accepted the legitimacy o f a

62

The development of social knowledge

policy that would allow children to be undressed. Before being presented with the hypothetical situation, most of the children had stated that it would be wrong to remove one's clothes in school. Nevertheless, in this domain of behavior, it was assumed that the policy could legitimately be determined by the school authorities. Sorne examples of children's responses to the questions posed in the Weston and Turiel study illustrate their use of the criterion judgments of institutional policy and authority power. The following excerpts, from a 5-year-old (David) and an 1 1 -year-old (Jane), w ere in response to each of the stories: David: (THIS IS A STORY ABOUT PARK SCHOOL. IN PARK SCHOOL THE CHILDREN ARE AL­

LOWEDTO HIT ANO PUSH OTHERS IFTHEY WANT. IT'S OKAYTO HIT ANDPUSH OTHERS. DO YOU THINK !T !S ALL RIGHT FOR PARK SCHOOL TO SAY CHILDREN CAN HIT ANO PUSH OTHERS IF THEY WANT TO?) No, it is not okay. (WHY NOT?) Because that is like making other people unhappy. You can hurt them that way. Jt hurts other people, hurting is not good. (MARK GOES TO PARK SCHOOL. TODAY IN SCHOOL HE WANTS 'rO SWING BUT HE FINOS THAT ALL THE SWINGS ARE BEING USED BY OTHER CHILDREN. SO HE DECIDES TO HIT ONE OF THE CHILDREN ANO TAKE THE SWING. IS IT OKA Y FOR MARK TO DO THAT?) No. Because he is hurting someone else. [Before specifying the school rule David was told about a boy who took his clothes off because he was warm from running around and asked if that was all right. ] No, because it's a school and other people don't like to see you without your clothes on. lt looks silly. (1 KNOW ABOUT ANOTHER SCHOOL IN A DlFFERENT ClTY; IT'S CALLEO GROVE SCHOOL . . . AT GROVE SCHOOL THE CHlLDREN ARE ALLOWED TO TAKE THEIR CLOTHES OFF IF THEY WANT TO. IS IT OKAY OR NOT OKAY FOR GROVE SCHOOL TO SAY CHILDREN CAN TAKE THEIR CLOTHES OFF IF THEY WANTTO?) Yes. Because that is the rule. (WHY CAN THEY HAVE THAT RULE?) lf that's what the boss wants to do, he can do that. (HOW COME?) Because he's the boss, he is in charge of the school. (BOB GOES TO GROVE SCHOOL. THIS IS A WARM DAY AT GROVE SCHOOL. HE HAS BEEN RUNNING IN THE PLAY AREA OUTSIDE ANO HE IS HOT SO HE DECIDES TO TAKE OFF HIS CLOTHES. lS ITOKAY FOR BOB TO DO THAT?) Y es, ifhe wants to he can because it is the rule. Jane: (IS IT ALL RIGHT FOR PARK SCHOOL TO SA Y CHILDREN CAN HIT ANO PUSH OTHERS?) No. You could break somebody's neck and nobody would care. You won't get in trouble, but

still I think that's not right. She might really get hurt, might even die and that pcrson might not even care. (SUE HITS A CHILD, ETC. !S IT OKAY FOR SUE TODO THAT?) No, she should wait her turn or ask someone else to ask that person. ·

[Before specifying the school rule Jane was asked if it was all right to go undressed in school . ] No, I don 't like streaking. 1t's weird, strange·. 1 just don't like it. (IS IT ALL RIGHT FOR GRO VE SCHOOL TO SAY CHILDREN CAN TAKE THEIR CLOTHES OFF lF THEY WANT TO?) 1 don't know. (IF YOUR SCHOOL SAID THAT, WHAT WOULD YOU THINK?) 1 wouldn't go to that school because 1 don't like to see everybody streaking. lt just isn't right. But if they want to, it's up to them. (BETTY TAKES HER CLOTHES OFF. !S IT OK FOR BETTY TO DO THAT?) 1t is really. They say you can, but 1 don't like it. (WHY?) Because 1 don't like to see people streaking. ·

David, the 5-year-old, expressed it clearly: Sorne rules should apply to all situations and the existing policy does not alter that assessment; other rules need

Dimensions of social judgments

·

63

not apply across situations. The justification for the belief that a school could not legitimately allow children to hit each other was the idea that it is not good to harm people ("It hurts other people, hurting is not good") . For David, therefore, the authority for moral prescriptions is neither in the dictates of specific persons nor in the group structure; for him the authority behind moral prescriptions comes from inferences that can be made about actions and their consequences to persons . By contrast, the authority behind conventions is in perseos and their positions in the social system. Furthermore, the children 's expectations of how an authority would respond to the permitted actions differed according to do­ main. Significantly more subjects expected teachers to contravene policy and reprimand a child who hit another than expected teachers to reprimand a child who took off his or her own clothes. Criterion judgments, justification categories, and familiarity with the issue

One of the conclusions that may be drawn from the studies just reviewed is that children begin to form social concepts at an early age, which is evident in their criterion judgments that are maintained as definitional features of domains across a wide age range. However, in conjunction with constancy across ages in crite­ rion judgments, age variations would be expected i n the organization of reason­ ing within a domain. Furthermore, the structural proposition that experience influences development implies that the children's previous familiarity with a task or issue should influence their judgments. Davidson, Turiel, and Black's (in press) investigation of criterion-judgment dimensions was combined with research about age-related changes in forms of reasoning and experiential sources of variation. There are three main differences between the Davidson et al. study and the others already described. First, severa! criterion-judgment dimensions were assessed for the same sub­ jects. The procedure was to present the subject with a brief hypothetical story and then pose a comprehensive set of questions pertaining to criterion judgments. (The types of questions are listed in Table 4.4, as well as five general criterion dimensions in which questions may be grouped for purposes of statistical anal y­ ses.) The dimensions tapped in the study included authority, punishment, rule alterability, and consensus.· To these are added two dimensions - subordinate jurisdiction and objective-subjective responsibility - that are directly related to propositions made by Piaget ( 1 932) regarding what he referred to as young children's heteronomous orientation to morality. Among other things, Piaget proposed that, in contrast with judgments made in late childhood and early adolescence, young children do not have an understanding of equality in peer relations and they judge by consequences (objective responsibility) rather than

64

The development of social knowledge

Dimensíons of social judgments

Table 4.4. Criterion judgment dimensions and questions Criterion judgment dimensions Subordinate jurisdiction

Domain

Familiar

Unfamiliar

3

46

Moral

37

o

3. The Do/lar. A student steals a dallar

Moral

14

authority? Can authority figure abolish this rule?

4. The Embezzler. A trusted employee

Moral

2

Is act all right, if punishment is expected? (if no) Is this because of punishment, or

5. The Scapegoat. A crowd is incensed at

Moral

5

Who do you think made this rule? Could the children have made this rule? Js it all right for authority figure to give/withhold punishment arbitrarily? (lf yes) Is this because of authority

though there is no one there to relieve him.

2. The Bul/y. A boy bullies a group of

children on the playground, pushing one off the top of a slide. which another boy had eamed b y doing chores for his mother.

role, or other reason? Is infraction all right if directed by

cleverly embezzles money from the company he works for.

other reason? !s act all right, if no punishment expected? (lf no) Is this because of absence of punishment, or other reason?

not being able to catch a thief, and

Is it all right to change this rule (unspecified change)? Is it all right to change an arbitrary­ detail of the rule? Is it all right to change the rule by group consensus? Is the act all right , if rule has been changed by group consensus?

Objective­ subjective responsibility

J . Physician's Duty. An emergency room surgeon leaves his post due to boredom al-

Moral

Acceptance of absolute authority power

generalizability

Story

Questions posed

Could the children get it changed?

Rule change and consensus/

. Table 4 . 5 . Descriptions of stories in the Davidson, Turiel, and Black srudy Frequency of use

over rule

Punishment orientation in evaluation of the act

65

Is act wrong if well-intentioned? Should actor be punished if well­ intentioned? Did well-intentioned or ill-intentioned actor do worse?

Source: Adapted from Davidson, Turiel, and Black (in press).

intentions (subjective responsibility). Findings pertaining to these two dimen­ sions are considered in Chapter 7 , where Piaget's theory and research on moral

1

punishes the man's innocent brother. 6. The Toy. A student takes advantage of his teacher' s trust and steals a toy belonging to the school.

Moral

7. Table Manners. A girl eats dinner with

Conv.

her fingers , knowíng this is not allowed in her family. 8. Marriage Surname. A couple gets married and the husband adopts the wife's sumame, contrary to custom. 9. Teacher' s Title. A student addresses his teacher by a first name, knowing that

o

43

Conv.

o

31

Conv.

11

19

Conv.

3

6

title (Mr., Mrs . , Ms.) plus s umame i s the required form. 10. Schoo/ Uniforms. A boy wears clothes of his own choosing in a school where students are expected to wear uniforms. 1 1 . Greetings. A girl greets her friends b y bowing rather than by the customary greeting. 12. Uniforms in Spom. A baseball player decides not to wear the team unifonn to a game, and wears a different color.

10

Conv. Conv.

3 2

o

judgments are discussed at Jength.

. The second difference between the Davidson et al. study and the others 1s that

age groups, each with 20 subjects: 6-, 8-, and 10-year-olds.) By using the pool of

assessments were made of the child's. previous familiarity with the moral or

stories and a standard way of assessing and coding the subject's familiarity with

moral or a conventlonal issue, were constructed. Sorne of the stories involved

domain and leve! of familiarity: familiar moral, unfamiliar moral, familiar con­

conventional event presented. A set of 12 stories, each dealing with either a

issues thought likely to be familiar to children, whereas others were designed so

as to present what were likely to be unfamiliar issues. (The study included three

the event, it was possible to obtain responses to four types of events combining

ventional, and unfamiliar conventional. A tab!e from the study containing

a

list

of the stories is reproduced here (see Table 4.5) to provide additional examples

66

of events classified as moral or conventional. The table includes summary de­ scriptions of the stories and the number of times they turned out to be familiar or unfamiliar events for the 6- to 1 0-year-old subj�cts. A third feature of the study is that in addition to assessments of criterion judgments, subjects were systematically probed for responses that could be ana­ lyzed for justification categories. A coding system describing the different cate­ gories used in reasoning about the events was formulated from analysis of a portion of the protocols and then applied to the remainder. Summary descriptions of the reasoning categories are presented in Table 4.6. Comparisons were drawn on both criterioo judgments and justification catego­ ries for domain and familiarity of event. The major findings were that responses differed in accordance with domain, that familiarity with the event had sorne influence - especially on the responses of the youngest children - and that there were age-related differences in justification categories. If the responses to famil­ iar moral and conventional events are considered alone, the judgments on the criterion dimensions are consistent with those in the other studies discussed. It was found that the authority and punishment dimensions were relevant to judg­ ments about conventional practices, but not to moral prescriptions. In the case of conventions, a person in authority (e . g . , parent, teacher, principal, police chief) could, in the judgments of the children, legitimately withhold punishment for an infraction, or even direct others to violate or eliminate a rule. Similarly, the absence of a punishment was judged to bear on the severity of an evaluation of conventional but not of moral transgressions. Children associate authority and punishment with the existing social organi­ zation. This implies that children recognize the possibility of other kinds of social arrangements, accepting that variations can coexist with regard to non­ moral issues. The children's responses to direct questions about the alterability of rules indicate that children do indeed comprehend variations in social arrange­ ments. When asked straightforward questions about whether a given rule could be changed, most children said that rules pertaining to conventional issues could be changed, but not those pertaining to moral issues. In turn, consensual agree­ ment was regarded as part of the conventional aspect of social organization. The influence of the child' s prior familiarity with an event was not as straight­ forward as might have been antícipated in that responses to familiar and unfamil­ iar events were not always different. The best way to summarize the findings on the familiarity variable is to say that the distinction between unfamiliar moral events and unfamiliar conventional events was not made as frequently by the youngest subjects, the 6-year-olds, as it was by the older subjects. Yet on sorne dimensions the 6-year-olds did respond similarly to unfamiliar events and to familiar ones. There is evidence, therefore, from the Davidson et al. study that children 's prior experiences do ha ve an influence on the formation of judgments.

67

Dimensions of social judgments

The development of social knowledge

Table 4.6. Justifica/ion categories Category

Descriptíon of responses includcd in category

Custom or

Appeal to personal and family customs ("Our family

Tradilion

doesn 't do il, so why should they?") as well as

social customs and traditions ("Because it's polite"; "Because you always do it when you rnanry") . Appeal to Authority

Appeal to the approval of specific authority figures

("If his coach gives him penn ission ,

l

guess

it's all right") or to the existence of rules ("Because if they make the n¡Jes you should abide by them"). Punishment

References to negative reactions of other persons

Avoidance

toward the actor, including social condemnation as well as explidt punishment ("lt's not okay because she could gel into trouble").

Prudential Reasons

Rcferences to nonsocial negati ve consequences to the actor, such as personal comfort or health

("Because it may be hot enough and so maybe he

could get sorne fresh air"; "Becausc if your hands

! 1 1

aren't clean, you might get genns or something"). Personal

Appeal to individual preferences or prerogatives

Choice

("S he should be able to say it; people who want to say one thing they can say it and people who want to say the other thing can say that"; "Because she thought it looked like fun").

Social Coordination

Appeal to the need for social organization or for

maintain i ng a system of shared expectations between

persons ("Because, another teacher of the same name, you'd get mixed up"; "Jf everybody took walks and not obeyed their jobs, then there'd be nothing get done").

Others' Welfare

A ppeal to the interests of persons other than the actor ("Because somebody could have gonen hurt, or you

know had to go to hospital or emergency"; "Nobody wants to get their money taken becausc they like to have lunch").

Appeal to Faimess

References to maintaining a balance of rights between persons ("l don't think it would be fair, where if someone eams some money and the other guy would take it and keep it for l:imself when he didn ' t do anything, he didn't earn it").

Obligation

References to feelings of obligation, including personal conscience ("Because my conscience would

1 ¡

bother me") as well as to personal duty ("You should keep your promises"; "He should do it for him be­

cause they are friends").

Source: Davidson, Turiel, and Black (in press).

68

The development of social knowledge

By the ages of 8 or 10 years, however, there is an increased ability to abstract elements from unfamiliar events, identify issues, and apply domain-specific judgments. Nevertheless, 6-year-olds did apply their judgments to unfamiliar events, though not as extensively as older children. It appears that experience influences conceptual development in conjunction with the child's ability to abstract from events and reflect upon them. These differences between the 6-year­ olds and the older children al so help explain sorne of the inconsistent results from studies on young children's judgments (see Pool, Shweder, & Much, in press). In severa! studies it was found that young children (3 to 6 years old) distinguished between morality and convention, but this distinction was not found in sorne other studies. lt is likely that variations in young children's familiarity with the events used i n the different studies account for the inconsistent results. The evidence from al! the studies discussed thus far is that, as children grow older, there is a fair amount of stability i n the ways they apply criterion judg­ ments to social events. Differing responses across ages, to sorne extent, are accounted for by the ways judgments are applied i n contexts unfamiliar to the child. Other age differences were accounted for by the categories used to justify courses of action, as is suggested by additional results from the Davidson et al. study. In their reasoning about moral issues, subjects most often used three of the categories listed in Table 4.6: welfare, fairness, and obligation. Moral prescrip­ tions were justified with reasons that appealed to the interests of others, avoid­ ance of harm, or a balance of rights and duties or obligations to others. The welfare category was the one used most frequently by children of al! ages. However, as illustrated by Figure 4. 1 , faimess and obligation responses increased with age for familiar events. The earlier appearance of an understanding of harm and welfare also manifests itself in an increased use of the welfare category by older children in their reasoning about unfamiliar events. The relationship between age and the justification categories associated with conventional events was notas clear-cut. The first six categories (and not welfare, faimess, or obligation) of Table 4.6 were associated with conventional issues. There were trends in the data indicating that the punishment categories decreased with age, whercas the prudential and personal-choice categories increased with age. Although this study did not provide adequate characterizations of changes in conventional reasoning categories , more extensive research on levels of devel­ opment is discussed in Chapter 7 . The Davidson et al. research does, however, point to both continuities (criterion judgments) and discontinuities (justification categories) in development. Discontinuity or continuity in the domain distinction

The evidence suggests separateness in moral and conventional thinking for sev­ era! aspects of both criterion judgments and justification categories. At the same

Dimensions of social judgments

69

4.5 o- _ - - -o WELFARE (UM)

4.0

3.5

3.0

t

0:: o u (1)

w

.J

¡:::: 4: o u.J z u.. o u.J o 4: fz u.J u

0:: u.J P.

40

40

20

20

o M

o

e

T

M

FIFrH GRADE

e

T

EIGHTH GRADE

100

1 00

80

80

60

60

40

40

20

20

justification category findings. The validity of conventional and task rules was judged, by subjects at three grade levels, to be part of the existing social system within the school. This is reflected in responses to four questions on the evalua­ tion of the act in conjunction with the presence or absence of rules. The results are summarized in Figure 5 . 1 . Figure 5 . 1 shows that, with regard to conventional

"O

o

o M

and task rules, most subjects stated that the relevant acts would not be wrong in

e

T

three situations: the absence of such a rule within their class, in a nearby school, lf-

_

•

These results, therefore, provide further confirmation of a distinction between morality and convention on the criterion dimensions of rule contingency and relativisrn. One question, however, did not serve to discriminate as clearly between the rule types. Most subjects (with the highest proportion in the second grade) rnaintained that it would not be right for another class within their school

Judgment of act in absence of rule

rule absence (al! the moral-conventional rules and moral-task rule comparisons reached statistical significance).

e

T

RULE TYPE

or in schools in another country. Conversely, just about all the subjects main­ tained that the moral acts would still be wrong in each of these three conditions of

M

e>- - -o

Evaluation of act in

absence of rule in another school

..._

Evaluation of absence of rule in another

cl ass

___...

Evaluation of act in absence of rule in

school in another country

Figure 5 . 1 . Percentage of subjects giving negative actand rule evaluations for moral (M), conventtanal (C), and task (T) rules. From Dodswonh-Rugani ( 1 982).

96

The development of social knowledge

to eliminate the conventional rules. A similar effect, though to a lesser extent, was obtained for task rules with the second- and fifth-grade subjects (see Figure 5 . 1 ) . The discrepancy between responses to this question and the other responses shown in Figure 5 . 1 provides a good example of what was referred to in Chapter 4 as a second-order phenomenon whereby, under certain circumstances, moral implications obtain for what is primarily a conventional issue. On the one hand, the act was judged to be rule contingent and relative to the social contexts of other schools. On the other hand, when the rule is applied differentially within the same social unit, issues of faimess and equity arise. The judgment that it would not be right to eliminate a conventional rule in another class represents a second-order phenomenon, because the primary justifi­ cation for the rule is nonmoral. Although the criterion judgments provide sorne evidence for this interpretation, further and more direct supporting evidence comes from the analyses of justification categories. In the Dodsworth-Rugani study, subjects' justifications of judgments about the rules were coded through a modified version of the scoring system used in the Davidson, Turiel, and Black study (the scoring categories were summarized in Table 4.6). One modification of the scoring system was the addition of a category labeled process concems, which were hypothesized to account for reasoning about the task rules. Process concems refer to a focus on work and learning activities and not on persons or social interactions. Another modification stemmed from one of the goals of the research, which was to determine whether understandings of conventional rules would be associated with concepts of social organization. Therefore, the analy­ ses of justification codings were grouped in four categories: ( 1 ) welfare and fairness (combining the others' welfare, appeal to faimess, and obligation cate­ gories), (2) social structure (combining custom or tradition, authority, and social coordination), (3) process, and (4) others (including prudence and personal choice). The percentage of responses coded in accordance with each of these four groupings, for all of the subjects, and summed across the different questions, are presented in Table 5 . 5 . Note that use of the welfare and faimess categories was infrequent for the conventional or task rules. However, insofar as those catego­ ries were used for conventional or task rules, it was usually in response to the question about the elimination of the rule in another class within the school. Also note that reasoning about conventional rules was almost always based on the social structure categories. The primary justification for the conventional rules was in social structure reasoning, but in a specific instance the rule could have moral implications - which is what i s meant by the second-order phenomenon label. The main import of the findings in Table 5 . 5 is the demonstration of the complementarity of criterion judgments and justification categories. Like the criterion judgments, the justification categories differed sharply (and with statis-

97

Rules and prohibitions Table 5 . 5 . Justificarion category responses for rule types (in percentages) Rule types Justification category groupings

Moral ( 1 )

Moral (2)

Welfare-fairness

94

90

4 o

2 o

2

8

Social stmcture Process Others

Conven-

Con ven-

tional ( 1 )

tional (2)

11

84 3 3

5

94 o

Task ( 1 ) 6 31

62

Task (2) 2 11

61

26

Source: Derived from Dodsworth-Rugani ( 1982).

relation does exist between domains and thinking about regulations in institu­ tional contexts. In turn, the findings can be viewed as evidence for the defini­ tional propositions set forth in Chapter 3 - namely, that conventions involve coordinations at the level of social organization and that the individual's concepts of social convention, unlike moral concepts, are closely related to more general concepts of social organization. Ascertaining that reasoning about conventional (but not moral) rules was subsumed under the categories grouped as social structure allowed for an analy­ sis of possible age-related changes in the organization of reasoning about conven­ tion and social structure. The analysis, which was based on reasoning specifically about the purposes and functions of conventional rules in the social organization of the classroom and the school, resulted in the identification of the five leveis summarized in Table 5.6. Dodsworth-Rugani found statistically significant dif­ ferences in levels of reasoning among the age groups (leve! l was associated with grade 2, leve! 3 with grade 5, and levels 3 and 4 with grade 8). As is indicated in Table 5.6, the levels identified in the Dodswoith-Rugani study characterize concepts of the integrative functions of rules for the social organization of the school. Age-related changes in concepts of social conven­ tions and social structure were also identified in previous research that did not focus on conceptions of school rules (Turiel, 1978a). Those characterizations of developmental changes, which correspond with the levels described in the Dodsworth-Rugani study, are considered in the next chapter.

The systematic nature of social rule concepts

In light of the research findings, sorne conclusions can be drawn regarding the concepts of rules held by children and adolescents. The evidence shows that from an early age children form systematic, but non-unitary, concepts of rules. Sup­ port was not obtained for the proposition that individuals have a unitary concept

98 Table 5 . 6 .

of rules, which develops from an earlier belief that rules are fixed, absolute

Concepts of integrative Juncrions of rules for social structure

dictates to a later understanding that rules have wlderlying purposes (and are therefore alterable by general agreement). When the domain of the rule is not

A . Rule functions to maintain social structure by controlling the behavior of individual; and defining

taken into account (and questions about rules are posed to children or adolescents

their role obligations.

Leve/ l.

The mle maintains the relationships between the elements of the social structure by

in the abstract), it may appear that younger children are more inclin